Refusing before session.receive left the packet unacknowledged: the client resent it forever and every later ordered write stalled. The test now uses the same peer and proves a later ordered write is admitted.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The hang was the fake server, not the writer: it is bound to one connection id, and two publishes in one test sent a second connect it rejected from inside the socket handler. The unacknowledged-send test also scripted an admit, whose reply piggybacks the put's ack.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultNetWriteNotSentException names the pre-send boundary (connect, hello, no endpoint,
or a send port that proves nothing was written). Every other forwarding failure keeps
the session claim. ForUnownedSchema and the routing refusal no longer carry exception text.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
An undecodable frame is classed legacy by Decode, so the schema-side guard was unreachable
and a receipt peer was never answered. The guard moves to the legacy decode.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Look the id up first and authorize only on first sighting; remember refusals within the cap; never re-forward an id whose primary stayed silent; end the connection when an answer cannot be queued; fixed refusal texts; limit_reached cap refusal; commit-point doc on ApplyPut/Delete.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Follows Heimdall's second review of #60: replay by (session,id) before any
authorization; refusals stored with their hash; delete shares the dedupe
table and settles capability; a refusal the queue cannot take ends the
session; a peer's refusal frames and a receipt session's refused deletes are
not violations; the write limit has its own fixed text; the sink trait says
Ok means committed and visible.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Opt-in per connection by the first hello (or first write); the server answers a
capable hello with its own. A put or delete commits, then gets write_admitted under
its id; every refusal is error.v1 with the id (an unreadable frame carries the id
when it still holds one). (actor, messageId) dedupe is bound to the content hash and
asked only after the write authority grants; same id other content is message_invalid.
The shard write forwarder says hello, waits for the primary's receipt, and the server
relays only what the primary answered under the original id.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Per Heimdall's review. The dedupe key lives for the life of its session, bounded per
session without forgetting an id the server promised to answer. A longer retention
is optional and never to be relied on. After a reconnect callers resend only writes
that are idempotent by record key and read back before resending any other.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A refused put, a full dedupe table and a reused id refuse the write, not the
session; only a protocol violation ends one. A frame that does not decode but
holds a messageId is refused with that id. The server answers a receipt-capable
hello with its own, listing both frames. The sink's authorization (new
authorize_raw_document, default admit) runs before the session's remembered
writes are consulted. A second hello does not change the capability: it is set
once, by the first hello or the first write. A receipt that cannot be queued, and
a write the payload budget would drop silently, end the session instead of
leaving the write unanswered. Refusals to a receipt-capable session are queued
reliably.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
cultnet-rs gains cultnet.write_admitted.v0, cultnet.error.v1 (v0 + messageId) and
write_content_hash. The server opts a session in only when its hello lists both frames
(framing only; admission rules are unchanged). On an opted-in session an admitted put is
answered with a receipt carrying its id after the sink accepts it, a refusal is an error.v1
naming the id, a resend of the same content is re-receipted without reaching the sink, and the
same id with other content is refused message_invalid (never the old receipt). Other sessions
are answered exactly as before.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
- messageId unique per authenticated actor and session; dedupe key (actor, id) bound to SHA-256 of
[operation, schemaId, recordKey, payload]; same id with other content is message_invalid, never the old receipt.
- A late answer after the window is recorded on the write (admitted_late / refused_late); no blind resend.
- A forwarder with no primary answer relays nothing; the client reports unconfirmed.
- RECEIPT_RESULT_WINDOW_MS 5000 (default), LEGACY_RESULT_WINDOW_MS 500 (fallback only), DEDUPE_RETENTION_MS as vector constants.
- The hello capability changes framing only and never authorizes.
- Batches can commit partially; the caller owns retry.
- Vectors: constants, worked content hashes, 7 dedupe cases, late-answer and window scenarios.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Specification and shared vectors only, no runtime code. A write is answered
exactly once with cultnet.write_admitted.v0 or cultnet.error.v1 carrying its
messageId; the new frames go only to connections whose hello lists them (today's
TS decoder rejects unknown keys and versions, so "old clients ignore it" is
false). Clients settle admitted | refused | unconfirmed; the result window is the
pre-receipt fallback and ends unconfirmed. Wire vectors, schema-invalid cases and
16 replay scenarios in write-receipt-vectors.json.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultNetWriteCompletion.SendAndAwaitResultAsync registers a refusal handler before
the send, then waits a result window (default 500 ms; CultMeshClientOptions.
SubmitResultWindow, CultNetSchemaWriteForwarderOptions.ResultWindow). A
cultnet.error.v0 inside it rejects with the new CultNetWriteRefusedException
(Code: write_not_authorized, message_invalid, ...) whatever its order against the
send or transport ack; a send returning no longer completes the write. The
forwarder's refusal now throws on the same path as an unreachable primary, so
CultNetDatabaseServer.TryForwardPutAsync/DeleteAsync no longer report a refused
forward as done. Raw send primitives (C# schema client/session, Python, Kotlin)
carry the doc line: fire-and-forget, an ack or a return is not admission, read the
result frame.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
One loop now serves both: every put acknowledged, then a refusal frame (typed
cultnet.error.v0) fails the call in either order against the ack, and a success
frame or the end of the new result_window option (default 500 ms) completes it.
The batch no longer waits on flush_reliable alone, so it cannot hide a refusal.
Raw send primitives (CultNetRudpSocketTransportConnection::send/send_reliable/
flush_reliable, tcp_framed send) carry the doc line: fire-and-forget, an ack or
a return is not admission, read the result frame.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The stub's 120 ms second packet fired after peer.close() and threw
ERR_SOCKET_DGRAM_NOT_RUNNING as an uncaught exception, failing whichever
publisher test was running when the timer landed (load-dependent). Timers are
now cleared in the test's after hook instead of swallowing the error.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
publishRudpDocumentOnce resolved when peer.flush returned (ack only) and read no frame, so a
refused put resolved. It now rejects with CultMeshRudpWriteRefusedError (the peer's typed
cultnet.error.v0: write_not_authorized, message_invalid) whether the refusal arrives before the ack,
after it, or within resultWindowMs (default 500 ms) after it; a success reply ends the wait early.
publishRudpDocumentAndWaitForReceipt rejects with the same typed error instead of timing out.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A Ping in the 2 s grace after the capped frame was answered with a Pong built from the
session's ack state, which includes the capped frame. Every reply while the session is
ending now zeroes its ack fields. The deadline no longer wraps at the end of the clock.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The schema server answered the 8th undecodable frame and ended the peer's session in the same
call; ending clears the session's pending reliable packets, so a lost datagram was never
resent. The session now stays until the peer has acknowledged every answer (a lost one is
resent by PollResends) or a 2 s grace runs out, serves nothing after the capped frame
meanwhile, and then sends the goodbye.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The capped (8th) refusal went out unreliable and unordered with the goodbye at once, so a
lost datagram left the peer without it. It is now queued like any response and resent until
the peer acknowledges it; the session ends then (or after a 2 s grace) and serves nothing
after the capped frame meanwhile. The answer, its resends and the poll that carried the
capped frame acknowledge nothing, as every refusal. A reliable queue that cannot take the
answer ends the session at once with the unreliable refusal, as before.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
endCappedSession deleted the session right after the answer was sent, so the resend
timer never saw it and a lost datagram left the peer without the answer. The session
now stays until the peer has acknowledged every reply (a lost one is resent) or a 2 s
grace runs out, serves nothing after the capped frame meanwhile, and then sends the goodbye.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The 8th refusal was sent unreliable and unordered with only the first response,
then the goodbye at once, so a lost datagram left the peer without it. All answers
now go the normal reliable path; the session ends when the peer has acknowledged
them (a lost one is resent) or after a 2 s grace, and serves nothing after the
capped frame meanwhile.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A map with an unknown schemaVersion was rethrown: unanswered, uncounted toward the
session cap. It is now answered message_invalid, reported, and counted like any other
unreadable frame, so the cap holds for it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Heimdall's parity ruling. The schema server ended a peer's session on the first
frame it could not decode. It now answers message_invalid (field "message") and
keeps the session; RudpCultNetSchemaServer.MaxConsecutiveRefusedFrames (8)
consecutive undecodable frames end it after the last is answered, and any
decoded frame resets the run. Handler faults still end the session at once.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Heimdall's parity ruling. A frame that is no msgpack map used to be dropped
silently; it is now answered message_invalid (field "message") and the session
kept. MAX_CONSECUTIVE_REFUSED_FRAMES (8) consecutive unreadable frames, or
consecutive deletes the write authority refuses, end the session after the
capped frame is answered. The authority's decision is unchanged.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Heimdall's parity ruling. The refusal is still answered and the session kept
below the cap; the frame that reaches CULTMESH_RUDP_DOCUMENT_SERVER_MAX_CONSECUTIVE_REFUSED_FRAMES
is answered too, then the session ends. The write-authority decision is unchanged.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Heimdall's parity ruling: the refusal is answered on the live session, and
MAX_CONSECUTIVE_REFUSED_FRAMES (8) consecutive undecodable frames (reset by
any decoded frame) or refused deletes (reset by any decoded non-delete) end
it. The sink gate is unchanged.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
An undecodable schema frame was dropped and a DocumentDelete ignored, both
silently. Now the first is refused message_invalid (details.field "message")
and the second write_not_authorized naming the record, each with a fixed
reason, ending the peer's session as the other refusals do. The sink gate is
unchanged. Both bytes are pinned against the shared error-vectors.json.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Q6-A of docs/cultnet-error-contract-cut.md. message_invalid joins the closed
cultnet.error.v0 code enum (both schema copies), with details.field naming the
request field at fault ("message" when the whole frame cannot be read).
C# ForMessageInvalid, Rust CultNetErrorCode::MessageInvalid, TS
cultNetMessageInvalidError and Python message_invalid_error all write the same
captured bytes (error-vectors.json plus the C# and Rust cases). cultmesh-py and
cultmesh-ts refuse malformed puts and deletes with it; the C# subscription
servers use it for the missing subscriptionId.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Review finding: NaN, Infinity or >2^31-1 becomes a 1 ms Node timer and would
refuse every write that is not instant; 0 and negatives were silently clamped.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A never-settling decide() held up the later frames of its session. The
server now races it against writeAuthorityTimeoutMs (default 5000) and
refuses with write_not_authorized on expiry. The granted-path test now
asserts the canonical schemaId the authority saw.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The C5 plan still said the cultcache-ts tag triggers the npm publish; F16 and the workflow say it does not. npm view 404 re-checked 2026-11-08.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A put with wrong-typed shard fields is judged with null shard fields first, then
refused as malformed once granted. The refusal's record key and reason are
clipped so it fits maxPayloadBytes. The allocation test watches transient
allocations; catch-path and delete tests assert ids, operation, reply count,
onError and handlers. README/CHANGELOG no longer claim the schema is resolved
after the gate.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A frame that fails to decode or parse is judged as the write it looks like, so
an ungranted peer gets write_not_authorized instead of silence; a granted peer
gets onError and a malformed_document_put/delete refusal. Adds raw-frame tests
(int, 2**70, oversized claimed length, non-decoding, non-map document) and
the timing residual note.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Deny by default with a grant-all opt-in; the gate runs before the put is sized,
resolved or decoded; refusal is the typed write_not_authorized. Adds
onDocumentDeleteRaw, refuses wrong-typed shard fields before the gate, grants
all writes in the existing server tests, and ports the cultmesh-py
remote-write-authority tests (no forwarding).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Review: trim() differs from string.IsNullOrWhiteSpace on U+0085 and U+FEFF; use the
package's isNullOrWhiteSpaceCSharp and pin both in the test.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
cultNetWriteNotAuthorizedError(recordKey, reason?) writes the C# ForRemoteWriteRefused
bytes (pinned to error-vectors.json). CultNetDocumentRegistry.canonicalSchemaId resolves
the id a put or delete writes under, from the id alone, for a server's write authority.
No behaviour change for existing callers; the registry already had a delete apply.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The RUDP document server sent a host sink's refusal as an untyped Error (code None).
It now sends write_not_authorized with details naming the record, as C#, TS and
Python do; the text stays the fixed variant text, never the sink's. The rejection
carries the put's record_key. Other rejections stay untyped. Unit test pins the
bytes to the error-vectors.json write_not_authorized vector; integration tests
check code/details on the wire and that a refused put reaches no sink.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultMeshLocalServer.start bound TCP to port 0 and then bound UDP to whatever number
the kernel picked, failing with EADDRINUSE whenever another process held that
number for UDP (this host has hundreds of UDP listeners), and left the accept
thread running on failure. Port 0 now retries with a fresh pair (bounded); both
sockets bind before any thread starts; an explicit port still fails honestly.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A non-string shardId or non-int shardEpoch (bool is not int) on a put or delete
is a malformed_document_put/delete before the authority is asked, so a
shard-scoped authority cannot be bypassed by a shard the store would file under
str(shardId). The authority receives the fields as sent.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
raw_put_target resolves only a put of the wire shape (string key, bytes payload),
so an int payload is never converted (bytes(10**9)) before the authority speaks.
Shard fields of the wrong type reach the authority as None. Tests: huge payload
refused without allocating, RUDP peer path (endpoint and transport), shard types.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Finishes the remote-write gate: exports, daemon --allow-peer-writes, the gate is
total (an ungranted peer learns nothing from malformed/unowned-put replies),
existing server tests grant explicitly, 12 tests port RemoteWriteAuthorityTests
(forwarding has no Python counterpart), refusal bytes pinned to error-vectors.json.
README documents the breaking change.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Adds write_authority.py, node.raw_put_target/raw_delete_target, the gate in
_handle_raw_put/_handle_raw_delete with the peer passed in, and serve_node/serve params.
Missing: __init__ exports and imports in facade.py/session.py, daemon --allow-peer-writes,
tests, README and Breaking note. Untested.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The header comment, semver-policy and F16 of the content map still listed cultcache-ts-v*
as a trigger. They now say publishing is deliberately off and that a tag run uses the
workflow at the tagged commit (cultcache-ts-v0.15.0 sits on 7d292085, before the guard).
semver-policy gains a "Before pushing a tag" note. The guard's comment moves above `tags:`:
inside the list it broke the three `workflow:` tests of check-changelog-semver.test.mjs,
which read `tags:` followed directly by its items.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Publishing @gamecult/cultcache-ts is deliberately disabled until ruled on, and whether
NPM_TOKEN is set is unknown. The other tag patterns and workflow_dispatch are unchanged.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
0.14.1 and 0.14.2 were never tagged; cultcache-ts-v0.14.0 is the last release. The
Unreleased exports (SchemaConflictError, schemaIdentityOf, the document payload codec)
are in no tagged release, which is what failed the declared-ranges check. Package and
lockfile go to 0.15.0 and every consumer range (cultnet-ts, cultmesh-ts, cultmesh-browser)
moves to ^0.15.0, since a 0.x caret does not admit a new minor.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
PR #20 made routingHint, code and details required on CultNetErrorMessage; six call sites in
cultmesh-ts still built the message without them, so build:ts failed with TS2345. cultnet-ts
now owns cultNetTextError(error), the legacy-union decode uses it too, and the call sites
and two test expectations call it. The types are unchanged.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Unscoped npm names are unclaimed and open to dependency confusion; cultcache-ts was
already @gamecult/cultcache-ts. package.json names, intra-repo dependencies, package-lock,
import specifiers (src, tests, samples, docs/start, first-call docs test), the workspace
list in run-typescript-workspaces, the verify-eve module map and package READMEs now use
the scoped names. Directory names, workspace paths in CI and release tag prefixes are
unchanged. Historical cut docs and past CHANGELOG entries keep the old names.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
load_soa installs a whole table and checks it with refuse_second_globals, but no test
reached that call (PR #15 covered put paths and pull_all). A table holding two records
of a global type is refused with CultCacheSecondGlobal and the prior view survives; a
table with one global loads and replaces the view.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Ruling (steward): add the code across C#, Rust, TypeScript and Python, each with a
captured-bytes case. details = { field: "schemaVersion", value: <offered version> }, no routingHint.
- schema (both copies), C# ForUnsupportedSchemaVersion + CultNetErrorCode.UnsupportedSchemaVersion in Rust,
TS code list, Python CULTNET_ERROR_CODES and schema catalog.
- bytes: error-vectors.json gains unsupported_schema_version (C# serializer output, checked by the
C# test, the Rust error_message.rs case, and the TS/Python vector tests; the Python builder reproduces them).
- cultnet-py interop_peer and cultmesh-py server now answer with the contract-shaped error (no extra
messageId, details {field,value}) via messages.unsupported_schema_version_error.
- tests that used unsupported_schema_version as the example of an unknown code now use not_a_code.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
cultmesh-portability's getting-started job checked out GameCult/Eve (404 on the forge) and
ran verify-eve-getting-started.ps1 -EveRoot ./Eve, which is Eve end to end. Nothing in it
runs without Eve, so the job goes. reactive-and-resource-lifetime has no Eve dependency
and is unchanged.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The package root imports node:fs, node:path and koffi through the single-file
store, so esbuild platform=browser fails on it. src/browser.ts is the root
surface minus that store; src/index.ts re-exports it and adds the store, so
the Node entry is unchanged. Exposed as ./browser and as the 'browser' export
condition on the root.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The sample ProjectReferences ../Eve and is not in CultLib.sln (PR #22).
README, getting-started 03/04 and browser-verse-transport now say so where they
tell readers to run it. Pages 01 and 03:73 only link Program.cs as source to
read, which needs nothing, so they are unchanged.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The sample ProjectReferences ../../../Eve, which is not checked out beside CultLib,
and the only error in the build was Program.cs CS0234 (GameCult.Eve missing).
The sample stays on disk and is built by scripts/verify-eve-browser-network.mjs
with --eve-root; it is just no longer part of the default solution build.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The forge runner is Ubuntu-only (Docker, one job), so Windows jobs queued on
push and pull_request can never run.
- cultnet-interop: the whole job is Windows-only; trigger is now workflow_dispatch.
- cultmesh-portability: both matrices keep windows-latest on workflow_dispatch
and run ubuntu-latest alone on push and pull_request.
- quic-native: unchanged; Build win32-x64 already runs only on dispatch or a tag.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Case one-version-rule runs version_ok/version_refused for notice_version, the sample id @N and terms_version; the Rust vector test handles both ops. README cites Heimdall PR #6 d6a8d42 for the rule.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
ECDSA P-256 route certificate and session proof verification ported from
CultMeshAuthorityProof.cs using the p256 crate (new dependency, pure Rust,
no default features), judged against the shared contracts/cultmesh vectors.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
ECDSA P-256 route certificate and session proof verification ported from
CultMeshAuthorityProof.cs, with typed refusals, judged against the shared
contracts/cultmesh authority route vectors (C# written and TS signed).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
check_version is the single rule; the sample id's @N and MemberTerms.terms_version both call it, so
terms_version "2026-11" is now refused and the golden vector carries 4294967295 (the harness now
validates the terms record). notice_version stays u32 and validate() refuses 0.
Tests: boundary table (1, 4294967295 accepted; 0, 4294967296, 4294967297, 01, +1, -1, empty, 2026-11
refused) across all three; each rule checked by mutation (u64 wrap, leading-zero, sign, terms unchecked,
returned value, empty sample, zero in both records) and killed.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Answers Heimdall's review of PR #12 (review 4): tests for the exact U+001F joined key and the empty and
separator refusals; an expect_terms vector op (newcomer case defends the MemberTerms pair) and
expect_no_terms; a mismatched member or experiment acceptance is never training data; the three scope
conditions kept, with a test of the rule over every scope pair; sample versions with a leading zero
are refused (README notes Heimdall may rule normalise instead).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Required is the contract's five keys (schemaVersion, error, routingHint, code, details), code is the
closed enum (or null), details and routingHint are object-or-null, and the catalog has no messageId
property, as cultnet.error.schema.json is closed. The test compares required, property names and the
code enum to the contract file instead of pinning the stale two-key form.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
tests/vectors/slot-nil was read by TS only. The Python reader accepts the same 15 rows
and refuses the same 2 as the C# reference and rewrites the reference's bytes.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
push, push_all and delete hold <path>.lock (flock; LockFileEx over the whole file on
Windows, as cultcache-ts store-file.ts) across the read and the replace. The staged file
is <path>.tmp-<pid>-<uuid>, fsynced before os.replace, with a directory fsync after off
Windows; a failed write removes its staged file and leaves the store.
Tests fail when the rule breaks: a write waits while a flock holder (Rust's lock) and the
TypeScript runtime's withStoreWriteLock hold the lock, three processes lose no writes,
fsync precedes replace, staged names differ, a failed replace leaves no staged file.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
pyproject (and cultnet-py) already declare msgpack as required; the code treated it
as optional behind a cultcache-py[msgpack] extra that does not exist.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
#[cultcache(name)], #[cultcache(index[ = alias])] and struct-level #[cultcache(global)]
in the DatabaseEntry derive; get_by_name, get_by_index and get_global on CultCache; a
second record of a global type under another key is a typed CultCacheSecondGlobal refusal
on every write path and on load.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Rulings: Heimdall PR #6 comment 759. Golden vectors for empty live text, empty revoked,
and received_at ties in both input orders.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Typed records, validate(), redacted([u8;32]) and the golden vectors follow
the nine rulings on Heimdall PR #6: u32 versions, 32-byte tombstone hash on
the exited Answer, scopeless replies stored with both fields absent, strict
training rule, fresh first reply per notice version. README cites doc lines.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
MessagePack encodes undefined as nil, and the peer rejects a document whose
sourceRuntimeId/sourceAgentId/sourceRole/tags are nil. Omit absent keys; test
that the encoded message carries none.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
NoticeAcceptance, Answer and MemberTerms cut from Heimdall PR #6
(docs/answer-collector-consent.md, head d836f29), with a golden vector
for versioned sample keys, scope-less exclusion from training, redaction
as hash-only tombstone and newcomer notice acceptance. Ambiguities are
listed in the README as questions for Heimdall.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Each sends a typed document as a document_put_raw message over a loopback
pipe and applies it to a second cache; a second test per runtime fails when
docs/start/README.md stops showing the snippet verbatim.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Authority holds one arena document in a CultCache and serves it over
CultMesh; two cultmesh-browser clients join, move and leave, and see each
other. Tested with real WebSockets. Browser run, persistence, remote trust
and CI wiring are marked not yet on the recipe page.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
README: the constant 1 in weight = 1 + log_b(amount/unit) is paid once per identity, so splitting
always pays until the log goes negative and no base fixes it; the defence lives outside the function.
Golden vectors (single and split k=1,2,10,100, three bases) computed with 60-digit decimal ln,
pinned by tests/golden.rs.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
weight = 1 + log_b(amount/unit) in integer cents and micro-weights with an integer-only
log, plus floor, per-identity cap and minimum amount per identity; splitting_gain/curve
report what splitting a budget over k identities does. 11 property tests.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Both blocks now come from a snippet region in their test, and a second test
fails when docs/start/README.md stops showing it, as the Rust and Python
blocks already do. The TS block differs only in the package import specifier.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Replace ascending-id trimming with pro-rata scaling by largest remainder
(ascending originator id breaks exact remainder ties only). Receipts carry
counted weight and the originator's latest covering revocation time.
Delegation::revoked becomes Option<u64>. Property tests: trim
order-independence, cap never exceeded, no delegator scaled more than another.
README: prior-art section, placeholders labelled.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Re-delegation to max depth, estate-of-delegator counting, per-estate cap on
the final holder, scope narrowing, direct override, revocation, uncast on
cycle/depth/trim, same-estate-only switch, per-person public receipts.
No stake function; weights are caller-supplied.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Pins that height rows built from math.phacelle (the C# mirror of cultmath_phacelle) are bit-identical
whatever order the rows are generated in, for a ring-buffer terrain. Golden fixture untouched.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Exact-case counts move: cycles_arcsine 154 -> 155 and cycles_arctan2 6 -> 7 on both routes; worst u1 and result hashes unchanged.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Each test file carries the snippet between markers; the docs show it verbatim and a second test in the file fails if the README block drifts. Matrix cells for Rust and Python CultCache now link them.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The glsl-parity.json header gains recorded_on (generator writes it; goldens untouched). CyclesOracleTests allows
cycles_arcsine, cycles_arccosine and cycles_arctan2 (glsl-parity families 95-97) 1 ulp from Blender, with the measured
cases named; a test pins the cap at 1 ulp, the three names, and that a 2 ulp miss still fails.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A language x package matrix (each cell a tested snippet or "not yet"), a C# first-call test and a TypeScript first-call test that the docs link, and a pointer from the top of README.md.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Recorded on Yggdrasil under llvmpipe (LLVM 20.1.2) with
packages/cultmath/tools/device/run-lavapipe.sh, WRITE_BOUNDS=1, at 990521b3.
Adds the lavapipe-glsl and lavapipe-hlsl rows for cultmath_fbm only. The
cycles_arcsine/arctan2 rows are left as committed (pre-existing asin/atan
oracle failures at base 94f20eee are out of scope).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
max singular value of the (cosine, sine) Jacobian <= max(10, 1.25 |side|) / (1 - normalization)
for normalization in [0, 0.9], |side| in [0.25, 25]; method and tightness in design.md.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
math.fbm(p, octaves, lacunarity, gain) and math.fbm(p) (four octaves, lacunarity 2, gain 0.5);
cultmath_fbm in the shaders, the GLSL regenerated by GlslLowering, a golden-fixture family
appended, FbmTests for the rules.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A write owns the record at its key when the written type declares the
record's schema id, or reads an older undeclared id of its own schema name
whole (ruling own-entry-reads-whole); every other registered type's record
is refused, in every cache, with or without a backing store
(cache-refuses-retype).
Removing a record and adding one of another type in a single batch is
staged as the held record's eviction then the add, so validators, Claim,
the store judges and the change journal see what they see staged apart
(one-batch-retype-is-eviction-then-add); Watch<old> gets Removed and
Watch<new> Added. A one-batch retype across home stores is refused by the
batch-spans rule and must be staged apart. A variant rebase changes type
exactly where a plain write may (variant-rebase-follows-plain-write).
Store kinds outside the library get one protected per-key judge
(store-kind-per-key-judge), exercised by tests/GameCult.Caching.OutsideStoreTests.
Spec variants:cut_spec:cut-write-ownership-cs.r1; reports h1-h5;
Soul s5 clears the merge (verdict variants:verdict:cut-write-ownership-cs.s5).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Merges hands/garden-path-remote-write-authority (51ca81a1..1e73e206).
cultmesh-garden-path:cut_report:cut-remote-write-authority.h1/h2; Soul s1 and s2 hold.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A real LiteNetLib pair on ephemeral loopback ports carries the handlers' answers to a client that decodes them with the
production schema decoder; no fixed port and no global frame provider. The delete under a schema alias and a null
decision are pinned where each is decided.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Mutation triage: removing the refusal exception's constructor body survived
because the handler threw on the uninitialized peer either way; the helper now
requires the failure to come from the peer's send, not from building the message.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Captured-bytes case in cultnet-rs error_message.rs, asserted byte-for-byte by
the C# schema contract test; the cultnet-distributed-database contract and the
Networking README state the remote write authority invariant.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CultNetDatabaseServerOptions.RemoteWriteAuthority is asked once per received
write, after the schema resolves and before the shard does; none configured
refuses every remote write; an authority that throws refuses and is logged.
A refusal applies, publishes, logs and forwards nothing. The host's own
CultNetDatabase.ApplyPutAsync/ApplyDeleteAsync ask no remote authority.
Existing server tests whose subject is not the grant pass a test-local
granting authority.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A second public door named like the peer path with no caller besides the
handlers. Hosts keep CultNetDatabase.ApplyPutAsync/ApplyDeleteAsync.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every fork the model page waited on is ruled, so its cells now carry the
rulings. GP1 confirms at runtime that the database server applies a put
from a peer that carries no identity. F13-F17 record what the remote
write authority cut stands on: peers have no identity, refusal codes are
a coordinated cross-runtime enum, forwarding re-sends as a new client,
the door has no consumer outside CultLib, and the Rust document server
has the same gap. The cut sequence splits authority collapse into its
remaining lines.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Probed at b7a68884: only the C# cache arm retypes a key (cache {A,B}
replaces B's record with A); cacheless writes already refuse. TS, Rust
and Python key records by (type, key), so two types coexist at one key
and no leaf has a retype path; leaf parity needs key identity first.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The package ships MessagePackAssumedFormattable(typeof(CultRecordRef<>)) as an
MSBuild AssemblyAttribute in buildTransitive, so generated resolvers defer the
ref to CultDocumentResolver. Geometry imports the targets and drops its three
per-member [MessagePackFormatter] workarounds. GeneratorConsumer is the
generator-on twin of ConsumerBuild.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Merges hands/variants-c2a at 7605bd17. Verified at 463714c4 (code tree
identical): every C#, TS, Rust and Python suite green on Yggdrasil and the
old-reader refusal gate refusing in all four runtimes
(variants:cut_report:cut-land-c2a-main.h1, Soul variants:verdict:cut-land-c2a-main.s1).
No tag: tags, Unity plugin rebuild and the declared-range gate are the 3.0.0 release cut.
Non-building branch commits exist; bisect with --first-parent.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The CHANGELOG bullet, the persistence-contract paragraph and the registry param docs of WriteRecord and WriteDocument now list the two ways a record is owned and every refusal the code makes, including a stored entry that lists another registered type's id ahead of the written one's and a call that writes no entry. The registry-less Shared read is described as it is today.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A cacheless call that writes nothing declares nothing and no written entry can claim the record, so the removal is refused and the file left as it was. Red under the Any-to-All mutant of the declared predicate in OwnsOnDisk, which owns every id when no entry is written.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The lent-shape test omitted the old id from the passed entry, so the base code refused it for the id and not for the shape. It now lists them, making the registry's wider entry the only thing that could own the record.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CatalogEntryLookup now takes the lent registry and answers from the written entries plus the registry's other types (descriptors whose schema id no written entry has). The registry Owns overload, the subset predicate and OwnsOnDisk's lent branch are deleted; the passed entry alone describes the written type, so a lent registry only adds claimants and can only refuse more.
Not yet built or run (verified by the job that follows).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The console logger repeated each line, so a raw count was 28 for 14 refusals. Names now prints the source (committed or live) and the gate counts distinct 'OLD-READER <runtime> refused <label>' lines.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The console logger repeats every test's output in its summary, so the log held 28 lines for 14 refusals.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A case that returned or passed early is Passed but prints no 'OLD-READER <runtime> refused' line. Each stage now requires the pinned number of those lines (14 C#, 3 Rust, 4 Python, 3 TypeScript) and fails '<stage>-refused-count' otherwise.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The cacheless written test now holds on the entry passed: the claimant's compatible ids must all be listed by the written entry of its id.
CHANGELOG and WriteRecord's registry doc describe the refusal through another registered type's listed id, the condition on own drift, and the Shared caveat. Not yet built or tested here.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
C# stages count TRX results with outcome Passed per class and fail '<stage>-notpassed' on any other
outcome (an [Ignore]d case is NotExecuted); Python also fails on skipped=, expected failures= and
unexpected successes=. Every stage prints its passed count. The committed 1.0.60 writer vector is
compared to a recorded sha256 (the 1.0.60 writer stamps wall-clock times, so no writer stage can
reproduce its bytes).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Kills the written-by-name mutant (the claimant condition compared schema names instead of ids).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Owns takes a condition on the claimant's answer (mayClaim): the cache arm accepts any
claimant, the cacheless arm only one of the entries the call writes. Another registered
type's record, under its current id or an old one it lists, is refused as at 0e57bf11.
Tests pin P1a/P1e/P1f shapes, the writer's own drift under a larger registry, and that a
second same-name type in the lent registry only refuses. CHANGELOG entry for the cacheless
claimant; stale comments on the store's owns predicate and WriteDocument's registry doc fixed.
Builds not yet run when committed; verified on Yggdrasil before push.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
summary_of ended in a grep that exits 1 on no match; under set -e that killed the script before it named the failed stage. With the
head ref-list test red (owns-declares-only) the gate exited 1 naming nothing; it now names head-ref-list and summary-diff.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The five TestCaseSource tests each run over two sources (committed, live): ten cases, plus the one OldReaderRefListProbe case. The
first run of the count check reported 10 and 1.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
OldReaderRefListProbe.cs is staged into the 2.0.1 tree by old-reader-refusal.sh and is not part of this tree's build; its method
lacked a closing brace at fd9bbaf4, which the gate's 2.0.1-reader stage reported.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
old-reader-refusal.sh pins the expected case count of each stage as a constant (C# stages from the TRX the run writes, by class; Rust,
Python and TypeScript from their own output) and fails naming '<stage>-count' on any other number, so a probe renamed out of its filter
or a skipped leaf test no longer passes with nothing run.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Only StoredAt moves (Soul s1 vec2.log); the writer is deterministic (two runs, identical trees). The directory page is renamed by its new
content hash. A README says how they are made and compared.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The 1.0.60 ref-list summary is taken after attach, an upsert of alpha, beta and the list reversed, and a flush, at 2.0.1 and at
head, and names records by schema id (the restamp-on-write ruling makes them equal). C2aGateWriter stamps every record with one
instant and compares the committed vectors byte for byte, the directory store as relative path and bytes with lock files left out.
The committed vectors are regenerated in the next commit, so TheCommittedVectorsAreWhatC2aWrites is red at this commit; it builds.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Kills the two survivors of the claimant lookup over written entries (Named inverted, Owner dropped).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
ReadStore refuses a record whose schema the catalog does not publish, so the
file-level scenario the spec named cannot be built; the rule is pinned where it is decided.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultDocumentRegistry.Claimant is now the only precedence that decides which entry
reads a persisted catalog entry, static over a lookup (a registry's indexes, or the
entries a registry-less cacheless write writes). ClaimantEntry is deleted. A cacheless
WriteDocument or WriteRecord lent a registry is never more permissive than a cache
registering the same types; an unpublished id is carried, not owned, through Owns' catch.
Fixes the four Low findings of variants:verdict:cut-restamp-on-write.s2.
Not yet verified on Yggdrasil when committed.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Hand mutants dropping each copy survived the full Caching suite; the flag is read only through the store's Entries (RefuseUndeclared), which
ToStoredDocument sets, so the cache's resolved variants and flattens were never asked.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Owns is now a static rule over (declared, claimantOf): a registry supplies both from its indexes, a cacheless writer supplies its one entry, so
the hand-written cacheless lambda is deleted and a renamed entry listing the writer's id is owned by both. The schema-name check is the
claimant selection and is pinned. Variant read-whole flag pinned for rewrite, flatten and remove.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Stryker found the true default and the two Wrote sets equivalent: a constructed or written record is stored under its type's declared id, so the flag is never asked. Not run at this commit.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The spec's account (guess null, then a successful attach) takes two gates, not three: the guess gate is taken only for a non-null guess.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The ForeignRecordTests read-only fixtures (another runtime's deck, the moving deck) had the same shape as the deck, which the
type now reads whole and owns; each gets a slot the deck lacks, so it is read only by dropping it. RestampOnWriteTests and the
successful-churn gate test are corrected from the first run. Not run at this commit.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultDocumentRegistry.Owns replaces Declares as the one answer to whether a writer may write, relabel or remove a record under a
schema id: the id is declared, or the type that reads it reads it whole (equal content hash, or compatible shapes with no persisted
slot ignored). Staging asks the declared half or CultStoredDocument.StoredSchemaReadWhole; the apply halves (single file,
directory, cacheless CultMesh) ask Owns over the durable catalog, once per distinct id. A record read only by dropping slots
stays read-only until its id is declared compatible.
Not built or run at this commit: verified on Yggdrasil at the cut's head.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultMesh was built for trusted co-op (Aetheria: Starbridge) with witness
authority deliberately deferred, so the current shape is the intended
first stage rather than a failed attempt at Byzantine tolerance.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Read-only survey of CultMesh's current witness path against crash-fault
consensus, CRDTs, BFT, game netcode and P2P anti-cheat, plus design notes
from the 2026-10-26 discussion: optimistic peer quorum with a server audit
slow path, network-proximity witness selection, dispute-scoped verification,
and cross-runtime determinism as a security property.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A cacheless write that re-enters its own store's hold now consults Cache first, so a journal's WriteDocument or WriteRecord gets the same refusal during an attach as after it (the attached text); the re-entry text no longer names an attach's pull. InStoreHold takes the store instead of its gate.
CacheBackingStore.GateTakes (internal, thread-static) counts the gates Gate takes; OneStoreCallTakesAtMostThreeGatesUnderFailingAttachChurn asserts at most three per call under failing-attach churn, which pins the stale-guess loop (M4b).
The contract and AddBackingStore say what holds: other threads may guess the tentative cache, but none runs a body against it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The folded-in writedoc case used to assert the refusal left the attach's registry in place and a later PullAll succeeded; the per-entry-point test now asserts that for every entry point.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The previous code commit's AJournalOnTheAttachingThreadCannotReachItsStore failed all 21 cases:
AddAsync mints its own key, so "later" never appeared. That commit builds; its new test is red.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Cache is the one attachment fact. AddBackingStore, the only writer, publishes it right after the
registry loan and sets it back to null in the catch, all inside its hold of both gates, so a
journal's store call during the attach is refused as it is after it (it used to stage or write
records the cache never admitted). Gate loses MaxPasses and its loop: the guessed cache's gate,
the store's gate (where Cache cannot change), then the settled cache's gate. Tests: one per store
entry point through a real attach (single-file and directory), each asserting the attaching cache
and a fresh cache agree after a flush; the second-store refusal pinned before the same-store one;
the three attach tests renamed to the published-cache shape. The old
AJournalOnTheAttachingThreadCannotWriteItsStoreCacheless is the writedoc case of the new test.
This commit builds; docs follow in the next commit.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Fixes two tests of the previous commit that failed on their own fixtures (the inner call's
exception was swallowed in the hook; a variant that shares its base's name is refused). Builds.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
InStoreHold gains a cacheless flag: on re-entry of the same gate a cacheless call throws
instead of running, so the attach's hold (until it publishes Cache) and an enclosing store
call's user code cannot write the store behind the cache. Gate passes the flag; no field
mirrors Cache. Tests through a real attach and a nested cacheless write, a header pin for
attached variant stores, contract and CHANGELOG.
Builds on the previous commit; neither is built or run until the verify job.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Deletes CacheBackingStore._codec and UseRegistry's codec parameter, the reflection LentCodec
test helper and the bare-store codec test. The cache re-resolves every pulled variant with its
own codec when it admits the load.
Not yet built or tested; the verify job at the head covers every commit.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
UseRegistry(registry, codec) is the single writer of both loans; the pull builds
variants with the store's own codec, a failed attach takes both back through the
same call, and AddBackingStore sets store.Cache last. Cache now goes from null to
a cache at most once, so Gate's retry makes at most two passes: the 1,000 stale
cap becomes MaxPasses = 2 and its comments say why.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Kills the survivors of the first mutation table (no recheck in the store branch, a Gate() that refuses the held cache, a stale
retry result, a same-store reentry refusal).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CultCache owns the per-thread gate facts: _storeHeld joins _held. InStoreHold marks the store gate; Gate() is where every
cache gate entry (holds and reads) refuses while that mark is set and the gate is not already the thread's. A getter run by a cacheless
call or an attach that reaches a cache or a second store throws instead of deadlocking against an attach.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Re-entrant getter during a cacheless write (read and write), a getter against the attaching cache, an attach pull that enters
a second store, and readers queued on a failed attach (snapshot, PullAll). Every thread join is bounded.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CacheBackingStore.Gate<T>(body, cacheless) reads Cache as a guess, takes that gate, checks the guess under it
and retries when stale. A cacheless call's refusal is the store-gate branch's validation outcome, in the same hold
as its body. WhileDetached becomes UnderStoreGate over CultCache.InStoreHold; AddBackingStore tests Cache, not Loaded.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
old-reader-refusal.sh gains gate <runtime> and per-runtime old revisions (b8fc5eb0 = cultlib-unity-v2.0.1,
02cf04b5 = cultcache-rs-v0.3.1, 45c2f400 = cultlib-unity-v1.0.60). C2aGateWriter writes the live v3 and v5 stores.
OldReaderGateProbe replaces OldReaderV3Probe. The vectors themselves follow.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The call no longer picks its gate from a read of Cache (Held), so a call queued on a cache's gate during an attach that then fails
cannot run beside a second cache's attach. Lock order: cache's gate then the store's, in the attach only.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Hand mutant T4 (lead >= 0x90) survived; the empty-array case now carries a newer stamp after it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Mutation of the restore line and of the header's Holds(runtime type) survived the previous tests (Stryker/hand mutants C1, C7). Not yet built.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The Unity list no longer calls MarkFlushSucceeded removed and places HeaderFor and the CacheBackingStore members where apicompat finds them. cultnet-rs gains a CHANGELOG. The store-composition contract moves DecodeDocument to the codec and states that a call's registry serves that call only. Docs only.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
SerializePayload takes the declared type. A call attaches its registry under the store's gate and restores the previous one, so no call leaves a registry on the store. Not yet run; the next commit set is verified on Yggdrasil.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The stamp functions (Rust refuse_unread_format, TS refuseUnreadFormat) are the single owner of the stamp rule, accept v1 and v3, and are called only from the one reader every whole-store write runs under the lock. The fix's header peek (TS refuseUnreadStoreBytes, peekHeader) is deleted. Its tests that v3 is refused and that a stampless file is healed are dropped; the reversal is a Breaking line in the Rust and TS CHANGELOGs. The 0xff-after-stamp test now runs over v1 and v3.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Mesh's typed single-file read and write call the store's typed members instead of planning ids through a public CultElementIds. Builds.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Stryker on WriteRecord left three NoCoverage statements (the two null guards, AttachRegistry) and a survivor (AcquireLock wait: true to false). Each now has a test that fails when it is mutated.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The lone-surrogate row now carries the value in the chunk hash and id slots, so removing text() there fails. Closes follow_up content-id-sanitiser-pin's pin.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Store-composition gains the Cacheless single-file access section; schema-compatibility states the slot-reuse consequence per variants:ruling:slot-reuse-documented. Owns follow_up slot-reuse-contract-doc.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Owns finding c2a-public-breaks-not-in-changelog. Leaf changelogs gain v3 read and write and element ids carried; cultcache-rs gets its first changelog (the whole-store writer fix is not on main, so no 0.3.1 section).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultMeshSingleFileDocuments mints, checks and counts element ids through CultElementIds, which the deleted GameCult.Mesh InternalsVisibleTo also granted; without the grant Mesh does not build. A cacheless typed writer has to plan ids before it serializes, so the three members it needs are public and nothing else of the class is.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
store_header and decode_grown both run through grown(); the copied maybe_grow and red-zone setup is deleted. Owns finding store-header-second-grown-decoder.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
DirectoryMessagePackBackingStore judges the parsed snapshot's stamp after the stamp-first read (an array with no stamp parses to v1 and was opened as a store). The NewerStampOverUnreadableBody rows assert the fragment each refusal names, plus a binary stamp over a valid body. DescribeHeader gets its suffix, digit-bound and zero-digit cases. Owns findings empty-array-directory-manifest-opens, stamp-first-unpinned-at-store-paths, describe-header-suffix-unpinned.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
SingleFileBackingStore gains the cacheless single-file read and write CultMesh used through internals; CultMeshSingleFileDocuments is now a caller of the public API only, and the GameCult.Mesh InternalsVisibleTo grant (with the duplicate Tests/MessagePack grants) is deleted. Owns finding mesh-ivt-moved-not-removed.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
refuseUnreadFormat is shared by isV1Snapshot and the writer's header peek; message text unchanged. Builds.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
One owner of the stamp rule, refuse_unread_format, serves read_all_unlocked and push_all. A file push_all cannot read is written as before. Builds.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Both new tests fail at this commit; it builds. Vector copied byte-identical from C2a (blob 9fc9b64f).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A staged write mints a later storedAt, so the 'rewrite kept storedAt' premise is gone; the pin is that a record rewritten under another schema fails Expect. The variant refusal names the renamed base through ChangedKeys, as C2a's typed exception carries it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Directory store now sets StoredSchemaId with StoredVersion after a page write (found by main's OwnPushOfALoadedDriftedEntryKeepsExpectHolding). RecordRewrittenUnderAnotherSchema stages the record, since a flush writes only the write set. Variant conflict test expects CultWriteConflictException.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Stamp rule runs on slot 0 before the body is scanned (JudgeStamp). DescribeHeader names a v<N> header that carries a suffix. Rust store_header walks on the grown stack. Drift tests declare Wide's id compatible (ownership by declaration). Mesh write-guard tests expect the typed refusals.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Identity: C2a's StoredSchemaId/StoredVersion owns; main's PersistedSchemaId, PersistedIdentity and both MarkWritten are deleted. Stamp rule: ReadStore judges the stamp before the body. Flush over damage refuses (merge-r1-precedence). Mesh single-file writes route through SingleFileMessagePackBackingStore. TS: store-file.ts commits, C2a's header handling kept. Rust: C2a's readable_store_format in front of the grown decoder. Build not yet verified at this commit.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Single-file PushAll reads the format stamp under the lock and refuses cultcache.store.v<N> with N > 2;
the directory store throws NotSupportedException for N > 4 from the manifest parse every write already runs.
An unreadable stamp keeps today's behaviour.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
C2a's StoredVersion owns the same identity and deletes both. No consumer outside CultLib uses them,
so they do not ship public in 2.0.0 and C2a's release is not a major.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The write replaced the file without reading it, so a 2.0 writer pointed at a store a newer
format stamped (or one holding other records) destroyed it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A protected static mark let any subclass mark another store's or cache's
entry. The instance form throws ArgumentException unless the entry is the
one this store holds in its own Entries at that key. Both library stores
already mark only their own entries, so their calls are unchanged.
Finding cultlib-gaps:finding:cut-expect-persisted-schema.s2.markwritten-unscoped.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The directory store's write of a generation marks each page it wrote once the
manifest names it, so a Push+PushAll of a loaded drifted entry no longer leaves
its persisted schema id stale. The single-file store marks through the same
member, and a store outside GameCult.Caching now has a supported way to mark
its own rewrite. The getter's doc names how a store records a rewrite.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Red until stores mark through a protected base member: the directory store's
Push+PushAll leaves the entry's persisted schema id stale, and a store outside
GameCult.Caching has no way to mark.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The directory store's pull filter compares the persisted schema id too; without a same-storedAt rewrite on that store nothing pinned it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The claim that every write mints a later storedAt is false for a whole-file write, which keeps storedAt and rewrites a drifted record under the writer's schema. State what the identity is and what a rewrite does to another cache's Expect.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The six reads that took Descriptor.SchemaId as a record's durable identity (ConditionsHold, Expect, the store-less commit, both pull filters, JudgeMerge) now read CultStoredDocument.PersistedSchemaId. The store sets it from the record it read, a stamped entry starts at its descriptor's id, and a completed whole-file write (single-file PushAll, whole-view CommitBatchCore) resets it under the hold. A variant and its resolved view share it. An unchanged record loaded under compatible drift now satisfies Expect and ExpectUnchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Red at b549d731: an unchanged drifted record fails Expect on both stores, and another writer's normalising flush is not seen. Pins the cases of cultlib-gaps:cut_spec:cut-expect-persisted-schema.r1.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Fixes variants:finding:cut-ts-store-write.s2.changelog-csharp-any-write-overclaim:
a C# conditional commit on a clean store lands onto the current file; other
C# writes can still overwrite concurrent TypeScript records.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
push and delete no longer lose records across concurrent TypeScript and Rust writers. A full
pushAll and any C# write still replace the whole store.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
cultcache-ts 0.14.1 re-exports checkPayloadDepth from the document entry on purpose, but the
closure script still asserted it undefined, so test:pack:ts was red and no workflow ran it. The
assertion now expects a function and the TypeScript job of cultmesh-portability.yml runs the gate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The default StoreFileOps were never observed. One test now wraps fs/promises.open and each handle's
sync and close, and records the store's bytes at every call: the staged file is synced and closed
while the store still holds its old bytes, and on Unix the directory is synced and closed after the
store changed. A second test puts a directory at <store>.lock and requires the lock to fail within
3 s with the lock path named, under the default 30 s wait bound.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A share-none open (what the C# lock and readers need) must succeed after the replace
and after a waiter gave up.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Both directions: a TypeScript write waits on a C# or Rust holder; a C# or Rust writer
waits on a TypeScript holder; the lock wait bound names the lock path.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each holds <file>.lock by its own runtime's protocol (share-none open; fs2 exclusive),
prints {"held":true}, sleeps --ms, releases.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
#enqueue takes withStoreWriteLock; the legacy heal moves out of pullAll's read into a
queued, locked operation that re-reads; push and delete read through #readDisk so they
never enqueue from inside the locked queue (a self-deadlock). Builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
commitStoreFile stages a unique temp, fsyncs it, replaces the store and fsyncs the
directory on Unix. Windows replaces POSIX-semantics first with a write-through
MoveFileExW fallback and a 25x20 ms retry on 5/32/33 (koffi, loaded lazily). The
cross-process lock functions land in the same file but are wired in a later commit.
package-lock.json also carries pre-existing version drift from regeneration.
Builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
hands/variants-c2a-element-id-holes at 64ec14f9 (cuts c2a-element-id-holes r6
and c2a-id-closed-world r2, held by Soul s5/s6 and closed-world s1/s2). C2a's
side since the fork is two prose-map commits; no code conflict.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Delete Hidden, HiddenProblem, UnreadKeyedMembers/UnreadMembers, Settle's
[MessagePackFormatter] refusals and the door's written-as-object branch.
Classify gains Leaf for attribute-formatted, [DataContract] and string-keyed
types; MembersOf honours the most derived AllowPrivate; the door refuses a
null StandardResolver formatter with the throwing path's text.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Probe facts from the id-holes Imagination passes (r5, r6) and Soul s5/s6;
the cut sequence names what landed since 2026-10-12 and what part 2 absorbs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
WrittenAsWalked required StandardResolver's formatter to equal one of the walk's own families, which refused every source-generated formatter. WrittenAgainstWalk asks one contrary resolver per walk kind and refuses only on a reference match. DcSeq is refused as a collection written as an object; FormattedRoles passes (registration owns the attribute).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
EnumeratesObjects took ElementTypeOf first, which for a dictionary is the
KeyValuePair (a leaf), so a keyed Dictionary subclass never reached the path; the
map's value type is asked first. The unreachable null-standard guard in
WrittenAsWalked is removed. Tests pin the direction and wording of the refusal, the
attribute-formatter collection case and the unwrapped inner exception.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultElementIds is the walk's model of a type; MessagePack's StandardResolver owns how
a type is written, and nothing checked that the two agree. ResolverClaim now asks,
for each id-path type, whether StandardResolver's formatter is the object or union
resolver's (a type read by key) or a collection resolver's (a collection), and refuses
the document at first encode or decode when it is not, naming the type. A keyed type
that enumerates objects goes on the path so a keyed List subclass is caught. A type
MessagePack cannot build a formatter for, and a resolver ahead that throws, are typed
refusals; the AllowPrivate catch is pinned (s4: keyed-enumerable-misclassified,
allowprivate-catch-unpinned).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
MembersOf is now MessagePack's StandardResolver write set: public fields and
properties with a public getter, none of them [IgnoreMember]. UnreadKeyedMembers
(renamed from NonPublicKeyedMembers) is every other [Key] member, including a
public property whose getter is private. IdProblem takes id candidates from every
instance member up the base chain, refuses an id MessagePack does not write, and
refuses a property id with a non-public setter, which MessagePack writes but does
not read back (s4: walk-member-set-differs, private-setter-id-dropped-on-read).
An oracle test pins the member set to MessagePack's ObjectSerializationInfo.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Classify tested IEnumerable<T> before [MessagePackObject], so a keyed class that
enumerates itself recursed until the stack overflowed during registration (s4,
keyed-enumerable-misclassified), and a keyed class that merely enumerates had its
keyed lists neither judged nor minted. Keyed first: MessagePack writes both by key.
VisitMemberType also stops at a type already on its own through chain, which ends
an unkeyed Node : IEnumerable<Node>.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
No fixture makes StandardResolverAllowPrivate throw when asked for a formatter (Stryker: the catch stays
NoCoverage), so the test pinned nothing the claim tests do not.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Stryker found the primitive, enum and object tests in Classify redundant with the value-type and unkeyed
branches, so they are gone. The new tests kill the survivors that were behaviour: a nullable keyed struct
walked as its struct, an unannotated struct treated as a value, [MessagePackObject] reaching a subclass,
and a claim on a type StandardResolverAllowPrivate cannot format.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
It writes the [Key] layout plus non-public keyed members, and registration now refuses any of those that
reaches an object list, so no element id can be dropped. The comparison is made lazily, only after an
ahead resolver returned a non-standard formatter, and an exception from it reads as not equal.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
MessagePack writes a non-public [Key] member only under AllowPrivate, and the element-id walk reads
public members, so the ids such a list holds could never be checked. Graph entries carry Hidden; a hidden
list or a hidden edge whose child reaches adds one problem and nothing to the id path.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
IsLeaf, IsObjectType and DictionaryValueType are deleted. Classify(declared) says Leaf, Map
(IDictionary<,> or IReadOnlyDictionary<,>), Sequence or Owner; a keyed struct is walked as an owner and
is never an element. Plan and Holds take the declared root type and follow declared member types, so a
member declared object or a subclass's extra members under a base-typed member are neither minted into
nor read. Minted path text is unchanged.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Graph edges and lists carry the collection types between member and child; Settle adds them to the path so
[MessagePackFormatter] on a collection class and a resolver claim on List<Elem> are judged. Declares replaces the four
MemberInfo.IsDefined member reads, which ignore inherit for properties. Survey.Judged dedupes the unkeyed-element refusal.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
_readHoldingIds and the pull's Holds walk are removed in the same commit that adds
SingleFileBackingStore.CopiedHoldIds: under the store's lock, while the header on disk is
unmarked, every plain copied record the writer's registry claims and whose type reaches an
object list is decoded, and the first holding an id marks the store. Positive-only, never
refuses a write. CultMesh's typed write attaches the registry its payload was encoded with;
the raw write attaches none. Contract text states the rule and its residual.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Stryker survivors on the id-path walk: dropping the union-subtype edge, dropping the
element base from the path. Both now have a test that fails without them.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The single-file store records, per pull, which plain records decoded holding a non-empty
element id and at what version. A write whose copied record is still at that version
marks the header (v3). The record is positive-only and version-gated, so nothing a load
records can unmark a store. The directory store passes false: only C# writes it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Registration: a concrete element type of an object list that is not keyed, and
[MessagePackFormatter] on any type or member on the id path or on the id member, are
refused. The id path is a fixed point over the walk's edges. The codec refuses, once per
document type and on every encode and decode path, a resolver ahead of StandardResolver
that formats a type on that path. UnionArmGuardTests moves StubArm out of the list union.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Written by PythonWrittenStoreVectorTests under CULTCACHE_WRITE_VECTORS=1 on Yggdrasil; not hand-edited.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Wrap and the Outcome arms gain id slots (unit arms become one-slot id records);
the Python unit-arm codec behaviour is kept on a separate unit union.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Written by the C# reference under CULTCACHE_WRITE_VECTORS=1 (NumericLeafVectorTests,
UnionParityVectorTests) on Yggdrasil; not hand-edited. Mutation testing does not
apply to regenerated vectors.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
LeafInner, the Node arms (Wrap new), the Outcome arms, and the GuardNode arms
(Branch, Unit8, StubArm new) take [Key] [CultElementId] string Id, with fixed
literal ids and hand-written emitters updated to the new slots. Goldens are
regenerated in the next commits.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Main's codec-document tests and the RUDP document-handle test read a record
stored under codec_note, plain_note or a foreign id through a handle for an
alias schema, and one store test picked among foreign-id records by whether
they parse. retire-schema-alias makes that read refuse; C2a's own tests
(a handle reads only records under an id its definition declares) pin the
rule. The codec reads and syncs under the definition's own id stay.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main's two payload-depth tests staged a record with no catalog entry and
asserted that the inspector and the single-file store infer its schema from
the payload's first element. C2a refuses a record whose schema id the catalog
lacks (merge-r1-precedence, exact binding), so the guess door is gone. The
inspector test keeps the bound check with a catalog entry that names the
schema; the store inference test goes with the door.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Main's documentFromPeerSnapshot takes a definition only, so the string-handle
read (first record at the key) that C2a held for Eve Electron cannot be
expressed; the assertion and requestCultNetRawSnapshotDocument's
firstRecordAtKey option went with it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
C2a's writer, catalog and soft-push tests handed raw Uint8Array payloads to
the store. Main made the envelope payload an opaque DocumentPayload, so they
wrap with documentPayloadFromBytes and read back with documentPayloadBytes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sixteen conflicted files resolved by the spec's rule: main owns payload
ownership, bounds and the Clock seam; C2a owns persistence semantics and exact
binding. MintStoredAt is internal static (previous, now); the cache and the
store pass their cache's Clock. Package versions are main's. The cultmesh-ts
payload-parse acceptance and its test are deleted under retire-schema-alias;
peer snapshot reads take the schema's own id, else an id the definition lists
as compatible. The held firstRecordAtKey fallback is removed: main's handles
take definitions only, so nothing could reach it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A non-home store (single-file and directory) is read only once the commit holds its own lock, so a foreign record landed at a checked key
while the commit waits is refused; and a directory judge whose lease is held makes TryCommit Contended at once.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CacheBackingStore.JudgeChecks (default: the view; single-file and directory stores read under their lock) replaces Land's view-only loop over non-home stores. Foreign takes the version the store read; CultForeignRecord.StoredVersion is get-only and set by the ctor, and every assignment after Foreign is deleted. One commit carries both spec commits 1 and 2 (the edits share hunks); unverified until the final job.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Remove and Commit of an unheld key no longer read the store at the call: the key travels as CultCommitRequest.Checks (commit) or a staged check (flush),
and each writable store judges it in RefuseUndeclaredOnDisk with the keys it writes, never removing it. A batch that only checks writes no file.
CultForeignRecord.StoredVersion records the version a store read, and the one ConditionsHold takes the store's foreign records.
Folds spec commits 1 and 2 (their hunks share the same methods). The src projects build; tests/GameCult.Caching.Tests does not at this commit
(CultCommitRequest gained a parameter; the next commit updates the test).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
gamecult_mesh_quic_native.dll.meta and msquic.dll.meta enabled the plugins
for Standalone: Win (32-bit) only, so 64-bit Windows players shipped without
the native QUIC bridge. Both now target Standalone: Win64 and are disabled
for 32-bit Windows. Managed plugin DLLs rebuilt by build-unity-package.ps1
-UpdateTemplate to carry 1.0.61; the native bridge keeps its 1.0.60 bytes
(its source is unchanged; an MSVC rebuild is not byte-reproducible).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ARefusalLeavesStagedWhatItDidNotRefuse re-staged the refused record and
expected the flush to refuse it; since b636a69f lists it foreign, that
re-staging is refused at once. The partial case now stages two records before
another writer makes one foreign.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A removal of a key the cache does not hold writes nothing, so it never reached
the apply; it was checked against the foreign list of the last load, and a
foreign record that arrived since was reported removed. Each store now reads
its durable state for such a removal (RefuseRemovalOfUndeclared) and refuses
through the apply's own RefuseUndeclaredOnDisk, which reloads and lists the
record. The load-time check (RefuseRemovalOfForeign) is deleted: the durable
read refuses everything it refused, and it could not save that read. A store
kind that cannot read its durable state keeps the view check as its default.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Refuse reloads a refused key as a pull loads it, and now that includes the
store's foreign list: a record another writer moved to an id this cache cannot
read is listed in ForeignRecords after the refusal, so the next write of that
key is refused at staging instead of again at the flush. Both store kinds give
Refuse the durable record's foreign entry (ForeignAt); the variant refusal too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A raw entry that lists the typed record's id compatible may replace it at the
original key, which is the only path left to the opaque-payload header rule.
Replaces RawPayloadWriteOverAMarkedFileKeepsItMarked, whose second-key case
duplicated the last step of the test above it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The persistence contract's rule 3 now says the store refuses, under its lock, a
write or removal at a key whose durable record is under an undeclared id, for
the store's own write API and for CultMesh single-file writes, which declare
through the catalog entry they write. The sentence that a declaring type stores
the record under its own id is corrected: Declares is registry-wide, so the
record lands under the writing type's id. The late-registration test is renamed
to match.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A record another writer moved to an undeclared id after the load, a staged
removal of one, and one that appeared at a written key are refused at the apply
and reloaded read-only. A direct store Push, Delete and CommitBatch of a
read-only record are refused and never relabel it. Removing a foreign record is
refused typed. A CultMesh write over a record its entry does not declare is
refused, and one whose entry lists the id lands under its own id.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Remove and a committed batch removal of a key the cache does not hold ask every
store whether it carries a foreign record there (RefuseUndeclared); one that
does refuses with the typed schema conflict instead of Remove returning false
and Commit returning Committed over a record the store still carries.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ForeignOnDisk and the two RefuseForeign calls in PushAll and CommitBatchCore are
gone. ApplyWriteSet refuses first, over every key it writes or removes, by
Registry.Declares for a cache and by the catalog entry it writes for a cacheless
CultMesh write. Both refusals reload through one DurableAt; the directory store
reloads through LoadPage, the per-record body of LoadRecordPages.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The held-only mutant of RefuseUndeclared survived the suite, because the
apply-time refusal still caught the write at flush. This pins the staging
half for foreign records in both store kinds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The persistence contract's relabel paragraph is replaced by ownership by
declaration: a record read through the catalog's schema name or a persisted
entry's list is read-only, and declaring its id claims it. CultForeignRecord
says no type can read it; the write APIs and MintElementIds name the refusal.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The relabel tests become refusals: ARecordReadByNameIsReadOnly (name branch)
and AnOlderVersionCannotWriteANewerRecordItReadsThroughTheNewerEntry (a
persisted entry's list). New: every write path refuses before staging,
declaring the id claims the record, a type registered after load claims it,
MintElementIds is refused by a read-only record, the undeclared dropped-slot
case, and a variant of a read-only base lands. The copy test also checks the
write beside the read-only record landed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Declares (registry) is the one answer to whether a cache may write, relabel
or remove a record stored under an id. RefuseUndeclared applies it from
Validate's Claim, which every cache write passes, to what the home store
holds at the key: a foreign record or one held under an undeclared id.
RefuseOverwrite and its two Push calls are deleted.
Red until rewritten: ForeignRecordTests.AWriteOfTheRecordItselfStoresItUnderTheTypesOwnId
and CompatibleSchemaIdTests.AnotherNewerCacheRestoringTheIdAfterAnOlderRewriteStillFailsTheCondition,
which pin the relabel rule 3 forbids.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The store chooses carried catalog slices by the entry object the read
produced. Keyed by schema id instead, a type whose id is unchanged but
whose entry changed (a compatible id added) would have the file's old
entry written over its own. The test stages a WsItem over a file whose
ItemId entry carries an unread slot and asserts the writer's encoding
lands, in both store kinds.
Closes variants:finding:cut-c2a-content-version.s3.catalog-writer-entry-unpinned.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Pins that a carried catalog entry keeps the store's slice, in the single-file and the directory store.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The C# peer gains edit-in-place (payload and override through the public snapshot
API; collection and nested object through the held document), and the four-peer
test has Rust and Python read each edit.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Delete RawBytes from CultPersistedRecord and CultSchemaCatalogEntry, with every
drop-on-assign hook and the catalog's content guard. A decoded record or snapshot
carries no stored bytes, so an in-place edit of its payload or overrides can never
be shadowed by cached bytes, and the public snapshot and record serializers encode
every record from its fields.
SingleFileBackingStore now reads through ReadStore, which returns the snapshot and,
parallel to it, the exact slice the file holds for each catalog entry and record. Its
durable index keeps, by record key, that slice and its SHA-256 (hashed when a
comparison asks), which is the record's version. A write re-encodes every record
the writer staged and copies every other record, and every catalog entry it keeps,
as the slice the file holds. The directory store keeps its manifest's catalog slices
the same way. SerializeSnapshot/DeserializeSnapshot hooks become
SerializeStore/ReadStore plus SerializeCatalogEntry.
A staged but unchanged non-canonical record is now re-encoded canonically and gets
a new version; an untouched one keeps its bytes and version.
Tests: in-place edits through the snapshot API (payload, override) and through a
held document (collection, nested object) are written as edited; an untouched
non-canonical record survives a neighbour's write byte for byte; a staged one is
re-encoded. The cost test counts record hashes.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The previous commit does not build: the widening helper read a byte string with an overload the test project cannot bind.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A record read from a store file carries the bytes it was read from (CultPersistedRecord.RawBytes, dropped when any field is
assigned) and is written as those bytes, the way a schema catalog entry already is. A version is therefore the SHA-256 of the
slice the file holds, whatever legal encoding its writer chose, not of C#'s re-encoding, and a write that copies a record leaves
its bytes and so its version alone. The contract says exactly that, so the parity cuts port it literally.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Adds CultStoredDocument.StoredVersion, CacheBackingStore.VersionOf and the
SingleFileBackingStore.SerializeRecord hook, and swaps every identity comparison
in the same commit: SameStoredIdentity stays the one comparison, over versions.
Deletes the Identity helper, the (schemaId, storedAt) Expected tuple, the
in-memory fake-record projection and the directory's storedAt arrival.
Breaking public API for 2.0.0: CultCommitRequest.Expected, ConditionsHold and
the abstract SingleFileBackingStore.SerializeRecord.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The foreign-record header test expected a marked header to drop while a record was copied
forward, which the header rule no longer does. A directory store loads a record from its
page, not the manifest, so its skew test makes both writers hold the record ahead instead of
editing the manifest.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A record's bytes cannot say whether it holds an element id (an id is an ordinary keyed
string), so the per-record fact decoded by whichever type loaded the record was a split
authority: a narrower type unmarked a v3 store over bytes holding ids. HeaderForWrite is now
a pure function of the durable header, the HoldsIds of the records the write encodes, and
whether it copies any record forward: a write that copies keeps a marked header, a write
that encodes every record decides by content. The directory store takes the same rule.
Also pins the three Low test findings deferred from c2a-write-set s3: the reload cause is
observed by type and message, the author's own Expect after a commit, and the directory
store's re-mint branches.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Ported from Soul's s2 probes: P5 skewed clock (equal storedAt re-mint, mint from the durable one),
D (directory store re-mint, as a store-kind parameter on the later-storedAt test), E (a disappeared
base refuses a variant write), P3 (foreign refusal whose reload fails carries the cause). New: a
foreign record a refusal reloaded is not seen as moved, and a writer that finds the file gone has
read nothing. ExpectOnAVariant is restored to pin that Expect tests only the stored delta: a stale
base reaches the write conflict, not a mismatch. Per-commit build not verified.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The directory store wrote an earlier storedAt over a later record it replaced, against the contract.
CacheBackingStore.StoredAtOver is now the one re-mint; the single-file apply and the directory
generation write both call it. A foreign refusal passes the reload's cause to
CultSchemaConflictException as its inner exception. Per-commit build not verified.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
ebd66e6f and cc8604ae do not build: the re-mint named MintStoredAt without its class.
Two tests pin the staleness rule for a write that would add a variant to a plain store.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The merge tests become the refuse, reload, retry rule; new tests pin copied records that
cannot be decoded, the later storedAt over a replaced record, a batch that removes a staged
key, and a Mesh write into a variant store. The persistence and composition contracts state
the rule for the parity cuts.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
ApplyWriteSet re-mints an upsert's storedAt when the durable record it replaces is not
earlier, so A-B-A on the same wall clock cannot read as unchanged. Two stale whole-view
comments are corrected.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CacheBackingStore.Refuse reloads the refused and moved keys through Loaded and only
then unstages them, or changes nothing; RefuseForeign calls it. A write into a store
that holds or would hold a variant lands only while the durable record headers equal
what this store last read (_lastRead), else CultWriteConflictException. SameStoredIdentity
is the one identity predicate.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A write decoded other writers' records on a write path. The decode is removed
here; the merge tests that depended on it go red until the refusal path and the
header-only staleness check arrive in the next commit.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CULTLIB_REAL_STORE names a copy of a real store: one staged write leaves every other record and
catalog entry byte for byte. New: a copied record changed after the load keeps a marked header; what
this cache wrote counts as known id-free; a batch wins over a staged write of its key; a write is
judged on the staged edits plus the variants another writer left in the file.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The unconditional-commit last-writer-wins test now keeps another writer's record; the load-minted ids
stay for MintElementIds after an unconditional write; a detached store marks the variant it writes, not
the one it copies; a variant write supplies the indexed override; the conditional/unconditional
comparison uses one batch (a conditional commit refuses staged writes by design).
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
New WriteSetTests (single-file and directory stores, flush and commit): another writer's write and
removal survive, a store that appeared after open is kept, an unstaged in-place change is not
written, a copied record keeps the catalog entry the store holds now, an entry's unknown trailing
fields survive byte for byte, conditional and unconditional commits write the same bytes, the
header drops only when every copied record is known id-free, v2 exactly with a variant. Mesh: a
write copies neighbours forward and replaces only its key. The lay-back and whole-view tests and
the flush-restamp tests are rewritten or removed with the rules they pinned.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
SingleFileBackingStore.ApplyWriteSet replaces WholeView and ReplaceDurable: under the lock, from the
durable snapshot read there, it removes staged removals, writes staged upserts and copies every other
record and the catalog entries they reference unchanged. PushAll, CommitBatchCore (conditional and
unconditional, one apply) and CultMesh's write call only it. Catalog entries read from a store keep
their exact bytes and are written verbatim while their decoded fields are unchanged. Derive no longer
lets a registered entry publish an id nothing staged sits under. HeaderForWrite is the one header rule
for the single-file and directory stores. A write is judged against variants another writer left in the
file. Tests of the deleted rules are rewritten in the next commit.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
_unowned with its rebuild, Owns, Publisher, WholeViewWrite, WholeView, SameStoredBytes and
WroteWholeView are deleted before the write-set apply that replaces them; PushAll and
CommitBatchCore still call them, so this commit does not build. The contract states the
write-set rule that the next commit implements.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A Starfire run showed the junction helper left its parent directory behind, which the "write went
through the link" assertion mistook for a write. The race test sees nothing-at-the-path windows on
Windows because File.Replace is not atomic to a lock-free reader; that is the store's replace and
is reported as a fork, so the test skips there with that reason.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Behavioural tests through Mesh and CultCache for each rule the cut owns, and the persistence
contract's single-readability paragraph states that one runtime has one reader, lock and replace
and that only nothing at the path is missing. Not yet run: verification follows in one job.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
SingleFileBackingStore gains ReadDurable and ReplaceDurable (internal): CultMesh's read and write
helpers reach the one reader, <file>.lock and the unique-temp atomic replace through them. ReadDisk
decides absence from the path's attributes, so a directory at the path is an IOException, not an
empty store. Includes the directory clause that the spec lists as a separate step.
Builds with the previous commit's deletions; the tree before this commit does not build.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Deletes CultMesh's ReadSingleFileSnapshot, HasOlderCatalogLayout, ReadLegacySchemaCatalogEntry,
WriteFileAtomically and the File.Exists gate, ReadStore's entry-reader delegate, and the three tests
of the older catalog layout. This commit does not build: the Mesh read and write sites still name
the deleted reader; the next commit routes them through the CultCache single-file store.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
cultnet-ts imports readSingleFileStore and cultmesh-ts imports schemaIdentityOf and SchemaConflictError, none of which cultcache-ts 0.14.0 exports. The new check builds each tagged release a declared range admits and requires every runtime name the dependent imports to be an export of it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
readStoreBytes reports an absent store only when the read and the lstat both find nothing. Any other lstat failure is thrown, and an entry the lstat finds that is not a link is reported as appearing during the read. The manifest walk also runs a soft pushAll over every vector.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Absent writes, a store it reads is left byte for byte, and anything the reader refuses (an unreadable file, a dangling link) is refused. The readability manifest walk now runs a soft flush over every vector, so TypeScript and Rust are pinned by the same files.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Opening a junction whose target is gone fails with UnauthorizedAccessException, not the
IOException the dangling-link rule promises. ReadDisk now turns that into the same
IOException. The dangling-link tests build a directory junction (mklink /J), which needs
no privilege, and the file-symlink cases skip where the account cannot create one.
The fixture removes links before its recursive delete.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The manifest held only three-byte and whole-store trailers, so a Rust reader that
tolerated a single trailing byte passed every suite while C#, TypeScript and Python
refused the file.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Mutation survivors of the authority reader (record count, schema id, type) and of the
reader's dangling-link error text each get a test that fails when the rule breaks.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
readAuthorityRecord decoded the store itself and accepted only cultcache.store.v1. It now calls
readSingleFileStore, so v3 reads, a zero-byte file is a StoreUnreadableError, and a refused
header is never echoed. Tests derive their bundles from the Rust-written fixtures.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
pushAll({soft}) asked readStore, so it skipped every file the reader refuses. It now asks
#readDisk: absent writes, readable skips, unreadable refuses. The reader is readSingleFileStore,
synchronous and exported, so a consumer reads a store through it instead of decoding one itself.
New tests: soft pushAll over absent, readable and five unreadable files; reading an absent store
writes nothing; a forged header that does not start at the prefix is never echoed.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The seam sat before the call, so a mutant that swallowed the real rename's result was invisible to it. It now makes the rename itself fail, on both platforms.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
store_header is the only reader of a store file's leading bytes. It returns the header slot (a MessagePack str only, never bin, as in C#, TS and Python) and separately the proof that the bytes are one whole array, so a newer store cut short after its header is still named unsupported. leading_header, read_array_header and read_string are deleted. decode_store_file takes the header the reader already found.
Vectors bin-header-v1/v3/v9 join the readability manifest as refuses x4. Tests: a rename fault step (WriteStep::Rename) shows a failed rename is NotReplaced, the store intact and no staging file left; every merging writer refuses through a dangling link; the unsupported-format refusal names v1 and v3 and echoes only ASCII-digit headers.
Not yet compiled or run at this commit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CultMesh's single-file reader had its own zero-length-means-empty branch, so a Mesh write
replaced a zero-byte file the cache refuses (merge-r1-precedence). It now asks the cache's
snapshot reader and gets its verdict. The auto-merged test that read a same-schema instance
through MeshNoteAliasDocument cannot be built under retire-schema-alias; it now states the
ownership rule it protected (a database serves only what a shard owns) with the one type.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Mutation survivors on the merge's refusal text (the two readable formats, the
unrecognised-header description, the record-slot refusal) had no test pinning them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
R1's rule that the header alone decides an unreadable-format store (a newer store
cut short is not damaged) is kept as leading_header, asked before the whole-file
proof. A directory at the store path now stops a write at the read as an I/O
error, so the two tests that used it to reach a failed rename assert that instead,
and the two c2a header tests assert R1's UnsupportedFormat kind, not c2a's old text.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The merge left store_header's visitor unclosed (does not build). R3's pin that a
whole-store flush replaces a dangling link survives the single reader: the reader
still refuses the link as an I/O error, and the flush, which has no content to
carry from a link that holds none, writes v1 over it (Rust write_all_unlocked, TS
pushAll).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
R1's rules win where both sides decided the same question: a zero-byte file is
not a store (refused as unreadable in every runtime), only nothing at the path is
empty, a dangling link is an I/O error, refusals never echo stored values and echo
a header only as cultcache.store.v<digits>, writes report Rejected, NotReplaced
and ReplacedNotDurable. C2a's rules win on what it added: the v3 element-id
header, variants, whole-file proof, typed unreadable errors in C#, TypeScript and
Python. Rust keeps R1's CultCacheStoreUnreadable{path, kind}; C2a's
StoreUnreadableError is deleted. The readability manifest's zero-byte row is
refuses x4.
Not yet built or tested: verification follows in the next commits.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A pull runs only on a clean store, and a store is clean exactly when nothing is staged, so the pull's clear was redundant and is gone.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CacheBackingStore.RefuseForeign now decides, for both store kinds, which written keys a foreign record refuses, drops the
store's staged change and the cache's copy of each through the Loaded path a pull uses, and throws one conflict naming every
refused key. The single-file store tracks the keys staged since its last load or write, as the directory store already did, so
a clean copy of a record another writer made foreign is carried, not refused.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The file's record moves when only its storedAt or only its schema id differs from the loaded one; a
staged removal is refused and dropped with the rest; a refusal leaves staged what it did not refuse.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The previous commit serialized every unowned record's document at load, which a variant cannot give
until the cache resolves it (five tests failed on it). A variant persists its delta, never its
resolved document, so there is no in-place change to detect.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
WriteGeneration left the refused key in the dirty and deleted sets, so every later generation
refused it. The refusal is now one check before any staging, shared by flush and commit, that
removes the refused keys from the pending sets. Tests cover both store kinds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A whole-view write laid back the loaded bytes of an unclaimed record over whatever the file held,
reverting its owner's newer write and moving storedAt backwards. It now lays back only while the
file holds the loaded (schemaId, storedAt); a newer record is carried as the file has it and a
removed one is not written back. The v3 header is decided by records that are written as the cache
holds them, so a laid-back id-less record no longer marks the store. A document changed in place is
written as a write of the record, not dropped. A refusal drops the store's copy of the refused
records, so it no longer refuses every later write.
Does not build with the directory store untouched? It builds; the directory twin follows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- The fixture gains an older entry listing the id ahead of the one that
owns it, and the whole-store write carries the record under the
registered type a Rust cache's pull gives it, not the entry's name.
Kills the own-id lookup flip and the ||-to-&& flip in the entry filter.
- A store path that cannot be read (a directory) is an error, not an empty
store (the NotFound guard had no test).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- The lay-back fixture gains an older entry listing the id ahead of the
one that owns it, and a test where another writer re-describes the id
after the load: a whole view (flush and commit) lays back the entry it
loaded. Kills the survivors in Publisher's own-id preference, the
laid-back entry's precedence in the written catalog, and the commit's
choice of that catalog.
- A store whose file is gone lists no foreign records.
- The registry's public resolution still refuses an unclaimed id, naming
the schema and the id (was NoCoverage).
- A directory load retried after a writer's commit lists each foreign
record once.
- Deleted LaidBack.ExceptWith(replaced): equivalent, since an upserted key's
new entry already carries what the cache would be told.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
C# writes a store; Rust, TypeScript and Python each write their own note
into it; the C# record and catalog entry are unchanged and C# reads the
store. C# writes again, and the other runtime's record and entry are
unchanged too. Compared on decoded MessagePack values, every slot of the
entry and its members included.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
C2a F2, widened by Self. The schema-identity probe showed a Rust write
replacing the C# catalog entry it read with a fabricated one: version
"sha256:...v1", members [], canonical JSON naming the hash.
- The store's one reader returns the file's catalog beside its envelopes.
- The encoder writes back the entry the file holds for each id it writes
(by own id, else a lister) and describes an id itself only when the file
has none, or when a registered type owns the id under another name (a
rename, whose rule is unchanged).
- Test: a push and a whole-store push both lay back a C#-shaped entry byte
for byte.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
C2a F2, widened by Self: records and catalog entries a runtime does not own
survive its writes byte for byte. The schema-identity probe showed a C#
rewrite restamping every other runtime's record to its sha256 id and
dropping their catalog entries.
- The single-file store remembers, per key, a record it loaded under an id
its type neither owns nor declares compatible (resolved through the
catalog's schema name or an entry pointing at a local id), with the entry
that published it. While the store still holds the entry it loaded, a
whole-view write (flush or unconditional commit) lays that record back
with its id, storedAt, bytes and catalog entry. A write of the record
itself still stores it under the type's own id.
- A record published by an entry under the type's own id is this schema's
and is re-encoded as before (the registered entry wins its own id).
- A laid-back record's load-minted element ids stay in memory only, so
MintElementIds still rewrites it.
- The rolling-deploy tests lose their premise: an older version's whole
view no longer sheds a newer record's member. They now pin that it lays
the record back and the newer cache's condition holds; the A-B-A test
keeps F6's defence through an explicit older write.
- Contract: the stamp and the lay-back rule in the persistence format.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The corrupt-store test pinned the old refusal of an unknown schema at open.
Under C2a F2 that record is foreign: the store opens, lists it, and opening
still rewrites nothing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
C2a F2, operator: "CultCache never destroys or relabels a record it doesn't
understand; keeps it byte-for-byte, refuses writes that would overwrite it."
- The registry's claim decision (Claimant) is split out of resolution. It
never decodes or compares shapes, so a store classifies a record from its
header. The last branch, "No local CultCache schema matches", now means
foreign instead of a refused load.
- Both store kinds keep a foreign record out of the cache and list it in
CacheBackingStore.ForeignRecords (key, schema id, catalog name, storedAt).
- Single-file store: a whole-view write lays back every foreign record the
file holds now, with its catalog entry, and keeps a marked header while it
carries one (it cannot see their ids). A staged write, a commit (onto the
file or whole view) and a whole view over a key the file holds as foreign
refuse with CultSchemaConflictException naming the key and id. A merge
judgement never decodes a foreign record, and one that replaced a held
record takes it away.
- Directory store: a foreign record's page is never read; a generation that
would replace or remove one refuses before any page is written.
- Contracts: hard rejection no longer lists "no compatible local schema
candidate"; the persistence format says what a foreign record is.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The handle named by a bare schema id yields unknown; 3c9 test code read .body
off it and did not compile.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Stryker survivors on 426c956a..fa266e58, each now a behaviour test:
- schemaIdentityOf's defaults (version "<name>.v1", compatible ids holding
only the own id) are asserted for a name-only definition.
- A lister bound twice is one lister: an unowned id it lists still
resolves to it rather than being refused as ambiguous.
- A snapshot filtered by an id a binding lists returns that binding's
records, on the typed and raw paths.
- The ambiguity refusal's record key and message tail are asserted, and a
raw snapshot record's payload encoding.
- A string-named store document reads the record under its id, and a
definition's own id wins over an id it lists wherever each record sits.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Soul F5. F7 (operator, 2026-09-30): "Schema aliasing sounds like a really
bad idea, I support retiring it." CultNetDocumentRegistry.Register wrote
both maps unconditionally, so the last binding for a schema id won.
- Register refuses a binding whose type already holds another schema id,
or whose schema id another type holds, with the reference runtime's
CultSchemaConflictException: the new binding's schema id, both document
types, an empty record key. Nothing is bound on refusal. Binding a type
to the schema id it holds again replaces the binding. This matches Rust
(fc3f3ee5) and Kotlin.
- Tests: both claimant shapes in both orders with two bindings ahead of
the holder (F6/M11), the first binding intact afterwards, and a same-type
rebind replacing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Soul F8 (M6, M7 survived): node.get, getRequired and put return the
object the cache holds, as C# and Python do (Rust copies). That is the
behaviour that ships.
- The README says the returned object belongs to the cache and must not be
mutated, and shows the put-a-new-value pattern.
- A test pins it: two reads return the same object, put resolves to that
object, and the caller's own value is never the one held.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Soul F7 (P9): a catalog holding two documents of one type answered a typed
lookup with the last one added, and nothing else could name the other. Two
documents of one type under different keys stay legal.
- CultMeshDocumentCatalog.tryDocumentById and documentById look a handle up
by its document id, the key the catalog already holds handles under
(it defaults to the record key).
- The publication-catalog test addresses both synced documents by id
instead of documents[0] and documents[1].
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Soul F3 and F4, and F1 in CultMesh's schema descriptor.
- The schema descriptor built from a definition uses schemaIdentityOf, so
a handle's schema id is the id the cache stamps.
- Store documents (F3): a handle reads the record under its schema id, or,
for a definition, under an id the definition lists as compatible, as the
cache resolves it. The catalog-entry name/version match and both parse
fallbacks are deleted: a record under another id is not this document,
whatever its payload parses as.
- Peer snapshots (F4): syncDocumentFromPeerSnapshot and definition-named
peer-snapshot handles take the peer's record at the key under the type's
schema id only. The shape filter is deleted and no match is not-found.
P8a (syncing A at k while the peer holds only B at k overwrote the local
B and then threw) and P8b (a peer's C stored as A) write nothing now.
- HELD: a peer-snapshot handle named by a bare schema-id string still takes
the first record at the key. Eve Electron reads the C# Aetheria daemon
that way (EveElectron src/cultmesh-provider-client.mjs:129-142): the
daemon stamps SHA-256 schema ids and the client asks by
"gamecult.eve.provider_advertisement.v1", so only the fallback finds its
records. How a TS reader names a C#-hosted schema is raised as a fork;
the fallback goes when it is ruled.
Tests: the store- and peer-snapshot tests that asserted shape routing are
replaced by tests of the refusal, including P8a and P8b, and one that pins
the held fallback.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Soul F1, F2 and F3 on the registry. The F2 principle (operator,
2026-09-30): "CultCache never destroys or relabels a record it doesn't
understand; keeps it byte-for-byte, refuses writes that would overwrite it."
- F1: the registry derives every binding's identity with cultcache-ts's
schemaIdentityOf, so it stamps, refuses and resolves exactly as the cache
does. P1 (A with schema name "p1.shared", B with schema id "p1.shared")
is now refused by both.
- F2a: an id nobody owns resolves to the one binding that lists it as
compatible. Several listers and no owner is a SchemaConflictError naming
the id and the listers, whatever order they were registered in; an owner
beats every lister. This is refused at resolution, as the cache refuses
it at load, not at registration: refusing at registration would make the
outcome depend on whether the owner was registered before the listers.
- F2b: version inference is deleted. "p3.note.v7" no longer resolves to
the type named "p3.note" and gets relabelled "p3.note.v1".
- F2c: a schema version string, and a schema name, are not ids. Snapshot
filters, typed puts and deletes, and raw puts match only declared ids.
- F3: a raw record under an id no binding declares is refused. The
registry no longer tries each binding's parser and stores the record
under the first that accepts it (P5: an identity parser registered first
took everything).
- The dead resolvePayloadSchemaVersion helper is gone.
Tests: the three shape- and alias-routing tests are replaced by P1, P2,
P3/P4 and P5 as behaviour tests. The one-type-per-schema test registers two
bindings ahead of the holder (F6/M11).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Soul F1 (hold on 98999997..6e47710e): the cache stamped records with
schemaId ?? schemaName ?? type, while the CultNet registry and CultMesh's
schema descriptor used schemaId ?? type. A definition that set only a
schema name was stamped one id by the cache and bound under another by the
registry, so its records could land on a peer as a different type.
- schemaIdentityOf(definition) returns the schema id, schema name, schema
version and every id the definition answers to (its own first). It is
exported, and it is the only place that derivation lives.
- The cache's catalog entry is built from it; nothing else changes here.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Stryker on 43ed9396..3763a53c left 20 survivors, in two places:
- cultMeshCatalogWriteArguments. No test wrote through a catalog's
replace or submitPrediction, so which argument was the context went
unobserved. A test now writes through both, in every calling form,
and checks the context and value that reached the document.
- The catalog refusal's schema names and message. The refusal test now
checks both names and that the message names both types.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
4002a996's refusal caught the self-test itself: the reconciliation block
registered kotlin.reactive_note beside kotlin.reactive_note.ui under one
schema id in one cache. The block tests reconciliation, not two types for one
schema, so its note is now kotlin.reconcile_note under its own schema id.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
43ed9396's type test caught it: TypeScript inferred the catalog value type
from a written value or an expected result through the conditional's false
branch, so replace and submitPrediction accepted any value and an annotated
collection read took the annotation as its type. The false branch is now
NoInfer, so only a definition or an explicit type argument decides it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Stryker left one survivor on 9adea366: emptying the refusal message. For a
same-type claimant the message did not name the schema id at all. It now
names the claimant's schema id in every case, and the refusal test checks
that it does. The message is built from document types and the schema id
only; registration carries no record, so no record value can reach it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
F7 (operator, 2026-09-30): "Schema aliasing sounds like a really bad idea,
I support retiring it." CultCache.register wrote its type and schema maps
unconditionally, and put, get, getAll and delete register implicitly, so one
Kotlin cache could hold two types for one schema and codecForSchema answered
with the last.
- register refuses a codec whose type already holds another schema id, or
whose schema id another type holds, with the new
CultSchemaConflictException (schema id, both document types, empty record
key), the Kotlin counterpart of the other runtimes' typed conflict. The
implicit registration in put, get, getAll and delete goes through the same
check. Registering the same type for the same schema again stays allowed.
- Self-test: other-type same-schema and same-type other-schema claimants,
through register, put, get, getAll and delete, each in a fresh cache; the
first codec still answers for its schema afterwards.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
F7 (operator, 2026-09-30): "Schema aliasing sounds like a really bad idea,
I support retiring it." CultNetDocumentRegistry::register inserted into its
type and schema-id maps unconditionally, so the last binding for a schema id
won, as the TypeScript registry did before 9adea366.
- register returns Result and refuses a binding whose type already holds
another schema id, or whose schema id another type holds, with
cultcache_rs::SchemaConflictError: the new binding's schema id, both
document types (a binding carries no schema name), an empty record key.
Nothing is bound on refusal. Binding a type to the schema id it holds
again replaces the binding.
- Call sites propagate the refusal: the cultmesh-rs registry macro, the
interop peer example, and the tests.
- Tests renamed to what they test: raw snapshot replication hydrates the
receiver's own type for the schema; a reactive document coalesces direct
member writes.
- New test: both claimant shapes in both orders, the first binding intact
afterwards, and a same-type rebind replacing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
F7 (operator, 2026-09-30): "Schema aliasing sounds like a really bad idea,
I support retiring it." A catalog indexed handles by schema id and by schema
name and version with plain map writes, so a handle of another type for a
held schema silently replaced it for every lookup by schema alone.
- Document and collection catalogs refuse, with the SchemaConflictError the
registries use, a typed handle whose type differs from the one holding its
schema id or schema name and version, or whose type already holds another
schema id. Nothing is indexed when a handle is refused.
- Allowed: handles of one type for one schema (one per record), the same
handle again, and a handle without a type, which claims none.
- Test: other-type same-id, other-type same-name-and-version and same-type
other-id claimants, both orders, via add and via construction, for both
catalogs; the first handle still answers afterwards.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The deleted parse option was the only thing that typed a catalog lookup, so
every caller had to name the value type. Catalog methods now take the schema
argument as a type parameter: a CultCache document definition yields its own
value type, an explicit type argument still wins, and a bare schema
descriptor yields unknown. CultMeshCatalogValue names that rule.
- Document catalog: tryDocument, document, latest, replace,
submitPrediction, authoritativeWriter, predictionWriter, watch.
Collection catalog: tryCollection, collection, latest, watchChanges.
- The two copies of the (value, options) / (context, value) argument split
collapse into cultMeshCatalogWriteArguments, and the typed lookups share one
private #require per catalog.
- Tests: the three reads that named their type infer it again. A type-level
test pins inference for reads, rows, watches, writes and writers with
@ts-expect-error on wrong assignments.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The catalogs' parse option went with asSchemaAlias, and with it the only
thing that inferred a catalog lookup's value type. Three test reads now pass
the note type explicitly, as a caller does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
F7 retires schema aliasing (one type presented as another for the same
schema). The Rust and Kotlin rows called their snapshot hydration "same-schema
alias" hydration. Both decode a raw record by its schema id into the one type
the receiving runtime registered for that schema; neither has a handle
aliasing API. The rows now say that.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
F7 (operator, 2026-09-30): "Schema aliasing sounds like a really bad idea,
I support retiring it." One type owns a schema, so a handle never needs
presenting as another type for the same schema. Parity with C# 1e045502.
- Delete asSchemaAlias from the document, bound document, collection and
bound collection handles, and cultMeshSchemasAreCompatible, which only it
and the catalogs' alias step used.
- Document and collection catalogs answer a typed lookup from that type's
handle or not at all. A descriptor without a type is still looked up by
schema id, then schema name and version. The parse option on every catalog
method existed only to convert an aliased value, and is deleted.
- CultMeshNode passes the caller's definition to its cache. The node no
longer resolves another type's definition by schema id, and no longer
re-parses values into the requested type, so a type the node did not
register is refused by the cache. node.get and node.put now return the
cache's own value, as CultCache.get and put do.
- Tests: the noteAliasDocument fixture is deleted. Tests that used it for
publication, sync, snapshot, store and prediction paths now use
noteDocument. otherTypeForNoteSchema stands for another type claiming the
note schema: catalogs do not find it, and a node refuses it on every path
and refuses to register it beside noteDocument.
- README stops documenting node aliasing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
F7 (operator, 2026-09-30): "Schema aliasing sounds like a really bad idea,
I support retiring it." CultNetDocumentRegistry.register silently replaced
the binding for a schema id, so two types could claim one schema and the
last one registered won.
- register refuses a binding whose definition claims a type, schema id or
schema name another bound definition already holds, with the
SchemaConflictError CultCache uses for the same refusal: the new
definition's schema id, both schema names, an empty record key.
- Binding the same definition again still replaces its binding, and a
definition listing another's id as compatible is not a second owner.
- Test: identical-id, same-name and same-type claimants refused in both
registration orders, leaving the first binding in place.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
F7 (operator, 2026-09-30): "Schema aliasing sounds like a really bad idea,
I support retiring it." C# now holds one type per schema, as TypeScript,
Python and Rust already do.
- Delete IsExactWireAlias and the branch that registered an identical second
type by CLR type only. A second type with the same schema name and version
is refused with CultSchemaConflictException naming both types and the
claimed id, whether it is identical, declares other compatible ids, or has
other members. The untyped DuplicateSchemaRegistration refusal and the
separate SchemaIdClaimedTwice path collapse into that one refusal.
- Owner-beats-lister (batch 10) is unchanged: one type owns an id, others may
list it as compatible.
- The per-name version index stays; it serves v1/v2 of one schema name, not
aliasing.
- Tests: the alias-registration tests are replaced by one refusal test over
identical, re-declared and re-shaped second types, in both registration
orders.
- cultcache-store-composition.md drops the AsSchemaAlias stale-protection note.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
F7 (operator, 2026-09-30): "Schema aliasing sounds like a really bad idea,
I support retiring it." One CLR type owns a schema, so a handle never needs
converting to another type for the same schema.
- Delete AsSchemaAlias from the document and collection handles and their
interfaces, with the MessagePack round-trip converters only it used.
- The document and collection catalogs look a typed handle up by its type.
The schema-name/version fallback existed only to find a handle of another
type and alias it.
- Tests: the MeshNoteAliasDocument fixture is deleted. Tests that used it
for publication, sync, snapshot and prediction paths now use
MeshNoteDocument; the one test that only checked alias refusal is deleted.
Catalog tests check a type with no handle is not found.
- public-api.md stops documenting aliasing; a stale comment in
ReadDocumentFromSnapshotResponse that cited the fixture is trimmed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Defends the id half of the whole-view restamp rule on its own: the record C# loaded
under v1's id is gone from the file when a v2 cache flushes it back, so there are
no stored bytes to compare, and only the changed id says it is a new store.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Owner beats lister, as in C# and Python. The cache kept one definition per id,
own or compatible, and refused a second with a plain Error, so a v1 definition
could not stay registered beside a v2 declaring v1's id (Soul F8, cult-cache.ts
registration). Now a definition owns its schema id and any number may list it:
a record resolves to the owner, else to the one lister. Several listers and no
owner name no single definition, and the load is refused with
SchemaConflictError naming the id, the record key and the listers.
Every registration refusal is now SchemaConflictError (Soul F10): a second owner
of one id, a second definition with one schema name, and a different definition
for one type. Each names the new definition's id and both schema names.
SchemaConflictError takes an optional message for these uses.
Tests: an owner and a lister register in both orders and each reads its own
records, and a lister alone reads the owner's id; two listers register in both
orders, a record under their id is refused typed without its owner and is the
owner's with it, registered before or after them; each of the three refusals is
a SchemaConflictError in both registration orders.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Owner beats lister, as in C#. A document owns its schema id; others may list it
as compatible beside its owner, in either registration order (batch 9 refused
any overlap, which forbade keeping v1 beside a v2 that declares v1's id; Soul
F8). A record resolves to the document owning its id, else to the one document
listing it. Several listers and no owner name no single document, and the load
is refused with SchemaConflictError naming the id, the record key and the
listers: refusing at registration would depend on whether the owner had
registered yet.
Every registration refusal is now SchemaConflictError (Soul F10): a second owner
of one schema id, a second document with one schema name, and a second
definition of one document type. Each names the new document's id and both
schema names and document types. Before, the last two raised CultCacheError.
Tests: an owner and a lister register in both orders and each reads its own
records, and a lister alone reads the owner's id; two listers register in both
orders, a record under their id is refused typed without its owner and is the
owner's with it, registered before or after them; two owners, one schema name
and one document type twice are each refused as a schema conflict.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Batch 9 refused any overlap between two types' ids, which forbade the migration
CompatibleSchemaIds exists for: v2 declares v1's id while v1's class stays
registered (Soul F8). The rule is now owner beats lister. A type owns its own
schema id; any number of types may list an id as compatible beside its owner,
registering in either order. A record under an id resolves to its owner, else to
the one type listing it. An id no type owns and several list names no single
type, so a record under it is refused, typed (CultSchemaConflictException naming
the id and the listers), where it resolves: refusing at registration would
depend on whether the owner had registered yet.
A second owner of one id is still refused at registration, typed. A schema id
hashes the schema's name and version, so the only second owner is a type of the
same name and version, which the name-and-version check already meets; the
separate own-id lookup could never fire and is gone.
Tests: an owner and a lister register in both orders, the owner's id resolves to
the owner, the lister's own id to the lister, and without the owner to the
lister; two listers register in both orders, a record under their id is refused
typed with no owner and resolves to the owner with one, registered before or
after them; a cache with v1 beside v2 reads each version's records as its own
type, in both registration orders. The emitted-fixture note says why a fixture
is emitted now that v1/v2 pairs no longer need hiding from the shared registry.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A condition names a record by (stored schema id, storedAt), and that name must
say which bytes are stored. A single-file whole-view write (plain flush, or an
unconditional commit) re-encoded every record under its own type's id with the
storedAt it loaded, so the pair could name two byte strings: v1 shed a member
keeping storedAt, a second v2 cache restored v2's id keeping it again, and a v2
cache holding the original bytes committed its condition (Soul F6, A-B-A).
Now such a write stores a record under a later storedAt whenever it stores it
differently: under another id than it was loaded under, or with other bytes than
the file holds at that storedAt (a record written with the ids it minted at
load, for one). A record stored exactly as the file holds it keeps its storedAt.
The store's entries and the cache's take the id and storedAt the write gave them.
Tests: the A-B-A through commits and through plain flushes, with Expect and
ExpectUnchanged; a flush that re-encodes a record restamps the cache that
flushed (a condition holds straight after, SM12); a record stored unchanged keeps
its storedAt through commits and flushes; persisting load-minted ids gives a
later storedAt. The rolling-deploy test asserted v1's rewrite kept storedAt; it
now asserts a later one, since that rewrite is a new store of the record.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A plain record and a variant loaded under a declared old id pass Expect and
ExpectUnchanged while the file is unchanged, and again after a whole-view commit
rewrites them (the variant case pins the id surviving resolution). A directory
store pull rereads a record whose manifest id alone changed, so a condition on
it holds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A directory-store record's identity on disk is its manifest entry, which is what
a commit condition compares; the page holds its own copy of the id. A load now
takes StoredSchemaId from the manifest, so a manifest rewritten under an older
id (AConditionOnARecordUnderADeclaredOldIdHoldsInTheDirectoryStore) is the id
the condition expects. Canonicalisation had hidden the page and manifest
disagreeing in that fixture.
The whole-view test's variant overrides its name: a variant may not share its
base's.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
register_document_type refuses a document whose own or compatible ids overlap
another registered document's, raising SchemaConflictError (empty record key,
both schema names, a message naming both document types), as TypeScript does.
Before, the record resolved to whichever document the dict held first (Soul F4).
With one carrier per id, resolution by id is a single pass. SchemaConflictError
takes an optional message for this use.
The test that pinned owner-over-lister resolution inside one cache becomes a
refusal test in both orders; shared compatible ids and a shared own id (neither
declaring its own) are refused in both orders.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Registration refuses a type any of whose ids (its own, or one it declares
compatible) another registered type already owns or declares, with
CultSchemaConflictException naming the id, both schema names and both CLR types.
TypeScript already refuses this (cult-cache.ts); before, C# kept whichever type
registered first (Soul F4). An exact wire alias must now also declare the same
compatible ids: a type differing only in its declaration had it dropped when it
registered second (H4), and is refused the same way.
Exact aliases that agree in every respect still register by type, and a record
under their shared id still resolves to the first registered (H3). Removing that
would delete CultMesh's same-schema aliases; it is left for an operator ruling.
Tests: both registration orders for a shared compatible id, an owned id declared
by another type, and a same-schema type with a different declaration; an alias
with the same declaration still registers. The registry tests use the shared
EmittedDocumentTypes helper in place of their own.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each stored document carries StoredSchemaId: the id its record sits under in its
store, the one it was loaded under until a write puts it under its type's own id.
Expect, ExpectUnchanged, the reload and merge identity checks compare
(StoredSchemaId, storedAt) with the store's bytes exactly. CanonicalSchemaId is
deleted: canonicalising the disk id let an older version's whole-view rewrite
(same storedAt, restamped to its own id, a member shed) pass a newer cache's
condition (Soul F1), and failed every condition on a record resolved by name or
through a file entry listing the local id (F3). A whole-view write marks every
record it wrote as stored under its type's id, in the store and the cache.
Tests: Soul's rolling-deploy probe as AWholeViewRewriteByAnOlderVersionFails...
(Mismatch for Expect and ExpectUnchanged, nothing derived lands, a pull shows the
rewrite and the retry commits); F3's H1, H1b, H2 pass Expect and ExpectUnchanged
when nothing changed; conditions after a whole-view commit hold for a plain
record and a variant. BareDeck is emitted at run time (EmittedDocumentTypes) so
the shared registry never sees it beside DeclaringDeck; the CanonicalSchemaId
tests and their Claimant fixtures go with the function.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A condition on a record under a declared old id is tested against the directory manifest too, and
a duplicate declared id is listed once. The merge-judging comparison keeps its plain schema-id
equality: a record it re-reads is only judged again, and no test reaches a difference.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The contract says the name a catalog entry carries is metadata and a schema renamed under a
stable id opens in every runtime that holds the id. The C# shape comparison refused any name
drift, so a record under a declared old id (the entry carrying another name and a different
content hash) or under a renamed stable id could not open. The comparison keeps the slot and
type checks; a record resolved by name has the same name by construction.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The C# interop note declares a compatible id and the peer gains a rewrite mode; the TypeScript
cross-runtime test has TS write a record under that id and another schema name, C# open and
rewrite it, and C#, TypeScript, Rust and Python open the result.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A type that failed to build a descriptor poisoned the shared registry's discovery of every
loaded document. A blank id is now filtered like the type's own id, and the test declares one.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultDocumentAttribute.CompatibleSchemaIds, as TypeScript and Python already name it. The
registered catalog entry lists the ids after its own, the registry resolves a record under one
to the type (its own id first, then the first registered type that declares it), and a
condition on a record compares schema ids through that resolution, so a record still under a
declared old id counts as the record the cache holds. The cache already stamps every record it
writes with the descriptor's own id; the store writers still change no identity, so a record a
commit does not rewrite stays under the id the file holds it under, published by the registered
entry. Without the declaration such a commit is refused, typed, as before.
Contract and XML docs describe the declaration.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The previous cut let any supplied type name a schema id over the arrived one, which let a raw
envelope of another type retype the records under an id (the dynamic-payload test refused it and
failed). A rename under a stable id is a write of the registered type that owns the id; a Rust
registered type's id is its entry type, so a supplied envelope whose type is its schema id names
the id and any other type under an occupied id is refused, typed, with the file left as it was.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The writer refuses a write that would change which type an existing record resolves to, unless
the entries of both the existing records and the written ones own the id (a rename; the registered
descriptor wins). Tests pin the arrived-overwrites-registered guard in both id orders, the
content-hash tie-break against the compatible-list order, sorted id processing and the cross-tier
refusal; a cross-runtime test has TS write a schema renamed under a stable id and TS, Rust and
Python open it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Deleted the heuristic that rewrote a record the cache held "unchanged since read" (same key and
storedAt) under the registered id, in the single-file commit and in the directory generation. It
could overwrite a record another writer changed while keeping its storedAt. A commit onto a file
now writes its batch and leaves every other record as the file holds it; a record under an id only
an arrived entry lists is refused, typed, with the file left as it was. Deleted the directory
store's _durableCatalog, a second stale-able source of arrived entries beside the manifest read
under the lease.
Tests: the lost-update probe, the two refusals that replace the restamp tests, the
arrived-overwrites-registered guard in both id orders, the content-hash tie-break against the
compatible-list order, and refusal determinism across record arrival order.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
pull_all_backing_stores resolves a record to a registered type by its schema id (a registered
type's id is its entry type) before the catalog name is tried. The writer tells the entries the
file already holds from the ones its caller supplies, and a supplied type names the schema id over
an arrived one, so a stable-id rename is not a conflict; two types of one tier under one id still
refuse the write.
Tests: a stable-id rename write, a renamed schema opening by id, and an id not overruled by a name.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A record now resolves to the document that owns its schema id, else one that lists it as
compatible, before the catalog name is tried; a schema renamed under a stable id opens, and an id
that names one document is not overruled by a name that names another. The name resolves a record
only when no local document has the id.
The writer refuses a write that would change which type an existing record resolves to, unless the
entries of both the existing records and the written ones own the id (a rename, the registered
descriptor wins).
Tests pin the arrived-overwrites-registered guard in both id orders, the content-hash tie-break
against the compatible-list order, sorted id processing, the cross-tier refusal and the rename, and
a cache whose document does not list the id a loaded record sits under.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Two entries of one tier that share an own id are refused when the id is carried, so a refusal at dedupe could never fire.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A read-back record's type is its schema's name and a supplied one's is the document type, so records of different tiers are never
compared; two supplied (or two arrived) records of different types under one id still refuse the write.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Every writer (C# single-file and directory, TypeScript, Python; Rust already did) chooses, for each schema id a record carries, one
entry that publishes it: the entry that owns the id (a registered descriptor, else an arrived entry), else one that lists it as
compatible. A registered descriptor wins over an arrived entry with the same own id. Entries are written as chosen, with no union of
compatible ids and no dependence on arrival order; an entry no record needs is not written. Two entries of one tier that share an own id
and disagree on the schema name, records of different types under one id, or a record no chosen entry publishes refuse the write with a
typed error (CultSchemaConflictException, SchemaConflictError) naming the id, the names and a record key, and the store is left as it was.
CultSchemaCatalogEntry.MergeById and the Python and TypeScript merge paths are gone. A cache-held record unchanged since the file was read
is rewritten under its registered id (C# single-file and directory), so an entry that lists an older id is not needed to publish it.
Changelogs for cultnet-py (removed export) and cultcache-py.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The directory store's preference for the entry being written was not pinned: nothing distinguished the entry's hash.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
cultnet-py's resolve_document_and_schema_id_for_raw_record is replaced by resolve_document_for_raw_record, which resolves the
document only; cultmesh-py keys previous values by the local schema id, so a snapshot under a compatible id or the schema name
reports an update with its previous value. Writers (C#, TypeScript, Python) merge same-id entries only when their schema names match;
another schema's entry is dropped or the record gets a default entry under its own id, and a record's own id is claimed before an
entry that lists it as compatible. The C# single-file writer prefers the entry being written now, as the directory writer does, so
name, content hash and members come from the registered schema and the compatible ids are the union. Vector:
compatible-before-owner-v3. A cultnet-py test's schema path no longer depends on the working directory.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The vector test alone did not distinguish the order on C#: the compatible entry's own list reached the same local schema.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Entries that share a schema id are one schema seen by different writers, so the writer lists every id any of them lists, in any
order: C# single-file and directory (CultSchemaCatalogEntry.MergeById), TypeScript and Python. Rust synthesises one entry per id and
has no such merge. An id that one entry owns and another lists as compatible names the owner in Rust, TypeScript, Python and the TS
inspector, as C# already did (own-id-over-compatible-v3). Also: a directory manifest in an unreadable format is the typed refusal
naming it, cultnet-py applies and reports a record under the local schema id, and Rust's error prints its cause once.
Vectors for the Mesh layout discriminator: a nil and an integer at catalog slot 5.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The two paths the previous batch left unpinned. Replication routing comments in cultnet-py and cultnet-ts say what the payload-based
and binding-based resolution is and is not.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The older-layout reader refused the second current-layout entry on its own, so an unconditional fallback still refused the vector.
With one entry, only a store not in the older layout is refused for its malformed slot, and the fallback's condition is pinned.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
cultnet-py resolves a raw record whose schema id no document lists by the schema version its payload opens with, and three
cultmesh-py tests (foreign UI-generated ids) depend on it, so that resolution is restored pending a ruling. The compatible-id
replication test goes through the real apply_raw_document_record and the reopened store.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The single-file and directory writers dropped an entry whose own id no record carried, leaving a record under a compatible id with
no catalog entry that publishes it: the same store nothing could open. An entry now stays while it publishes any id a record
carries. The conditional-commit readability test over compatible-id-only-v3 pins it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
put_envelope, put_envelopes and load stamp the registered schema's id, and the store writer publishes every record's id, so a
replicated record under a compatible id no longer leaves a store nothing can open. cultnet-py resolves a raw record by the schema
ids its documents list and nothing else.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
putEnvelope and admission stamp the registered schema's id, and the store writer publishes every record's id (the entry the
envelope carries when it lists that id, else a default), so a replicated record under a compatible id no longer leaves a store
nothing can open. cultnet-ts stops preserving the incoming id. The inspector refuses a store of the wrong slot count and a record
whose schema the catalog does not publish, instead of recovering one from the payload.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A record whose schema id the catalog lists only as a compatible id now opens (the cache resolved a persisted schema by entry id
alone). CultMesh's fallback to the older catalog-entry reader fires only for a store in that layout, decided by the entry's slot 5,
not for any decode failure, so a current-layout store with a malformed slot is refused as CultCache refuses it.
CultStoreUnreadableException.Path names the refused file. New vectors: compatible-id-only-v3 (accepted by all four) and three
current-layout catalogs with one malformed slot. The contract states the writer's invariant.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Rust, TypeScript and Python each had a test that a record missing its catalog entry is recovered from its payload; each now asserts
the store is unreadable.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The persistence contract requires the catalog to publish every schema its records name and specifies no recovery, so the catalog
is the store's own description: Rust, TypeScript and Python stop inferring a schema name from a payload's first field, as C#
never did, and a store naming an unpublished schema is unreadable in all four. missing-schema-versioned-v3 pins it with payloads
that open with a schema version.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
variant-slot-v3 overrides the variant's name, which the reader requires of a variant, so C# reads it and the others refuse the
fifth slot. The legacy-layout catalog vectors carry an entry per schema, so a store the reader accepts differs from the one it
refuses only by the bytes after it. missing-schema-v3 is refused by C# too.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
C# read such a store on flush and commit and refused it on open, with an untyped InvalidOperationException from the cache. The
store reader now refuses it with CultStoreUnreadableException, so open, flush and commit give one verdict, as Rust, TypeScript
and Python already do. The readability open test builds a cache with a codec, so a variant store opens as it would in use.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The refusal is a ValueError subclass with the cause chained. A store with more than three slots is refused instead of read.
The readability test walks every vector in the folder, asserts the error type and cause, the header a replaced file carries
and what it holds.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
pushAll, push and delete read the file through the one store reader that pullAll uses, so a store holding a variant or a bad
body is not replaced by a header-only check. A store with fewer or more than three slots is refused. The readability test
walks every vector in the folder, asserts the error type and cause, the header a replaced file carries and what it holds.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
push_all and every rewrite read the file through the one store reader (framing, header, schema catalog, records), so a v3
store holding a variant or a bad body is no longer replaced by a header-only check. The refusal is one typed error with the
cause as its source. The readability test walks every vector in the folder, asserts the error type and the header a replaced
file carries, and the truncated v1 and v2 stores are refused again.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A store refused for its framing, slot count or a record keeps that refusal instead of being re-read with the older layout, so
the record it names is the record the message names.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultStoreUnreadableException (a NotSupportedException) is the one refusal for a store file this runtime cannot read, thrown
by the single-file reader, the rewriting paths and CultMesh's reads, with the cause as the inner exception. The reader is exactly
one array of three slots (or an empty array): a header alone and an extra top-level slot are refused instead of read as empty or
skipped. CultMesh's older-catalog fallback is the same reader with the older entry reader, so it also requires one array.
Vectors: bad-body, header-only, missing-schema, variant-slot, extra-top-slot in the readability set, and a legacy-catalog store
with and without bytes after it for CultMesh. The persistence contract says what is true.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Rust's read path let a legacy envelope array with trailing bytes through when the
whole-array check was loosened, and no vector caught it. Add the vector; every
runtime refuses it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
AddBackingStore pulls, so a file the reader refuses throws on attach, outside
the assertion. Wrap attach and pull together.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The raw-write refusal test spelled the old header-only verdict (NotSupported for
a first slot that is a string, a legacy array replaced). The shared vector test
now drives both write paths; the old test keeps only the marked-file case.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
A rewrite replaces a file only when the runtime's own reader opens it. C#
drops ReadStoreHeader (a second, header-only verdict that let a flush replace
files the reader refuses) and asks DeserializeSnapshot from open, flush and
commit; the reader now requires exactly one array, so C# refuses bytes after
the store on read as TS and Python do. A zero-byte file is an empty store for
all three paths. CultMesh.WriteSingleFileDocument reads the file it replaces
through the Mesh reader before writing. Rust reads and rewrites through one
classification (trailing bytes refused on read; a non-store first slot must be a
legacy envelope array the reader accepts). TypeScript checks a legacy array with
pullAll's own decoder before replacing it. Python already conformed.
Shared byte vectors and per-runtime verdicts live in
tests/vectors/document-variants-c2a/readability; every runtime's test reads
them. Holds gains second-dictionary-value and second-list-item cases. The
persistence format contract now marks v3 by content and states the readability
rule.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
CultElementIds.Holds is the one fact a store marks its header by; a test now
drives it directly over a dictionary value, a list of lists, a list under an
unset element and an element, and over unset ids that hold nothing, so a
mutant that skips any of those branches fails.
C# ReadStoreHeader refused a first-slot string without the cultcache.store.
prefix and then ran RequireSingleFileFormat, which refuses every string that
is not v1, v2 or v3 anyway. The prefix check is gone; a non-header string is
refused as a format this runtime does not read.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Soul's mutant deleting the empty-file arm of write_all_unlocked survived: no
test rewrote a zero-byte file. Without the arm the empty file reaches
store_header and is refused as truncated. The test pushes onto a zero-byte
file and expects a v1 store holding the pushed record.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A rewrite read the header of the file it replaces but not whether the file was
a store: C# ReadStoreHeader skipped one value and ignored bytes after it, and
took a non-array value or an array whose first slot is a string that is not a
store header for a headerless legacy file (so did the C# Mesh raw write, which
uses it); TypeScript's pushAll took any non-array value or non-header first slot
for v1; Rust ignored bytes after the array.
The rule, one per runtime: a file is a store on rewrite only when it is exactly
one complete MessagePack array. Bytes after it, a value that is not an array,
and a first slot that is a string but not a header this runtime reads are
refused, and the file is left untouched. An empty file and a legacy envelope
array (first slot not a string, empty included) carry no header and are still
rewritten.
- C#: ReadStoreHeader refuses non-array values, trailing bytes and a non-header
first string, with a NotSupportedException naming which.
- TypeScript: storeHeader (store-format.ts) decides the rewrite header;
@msgpack/msgpack's decode already refuses trailing bytes.
- Rust: store_header(bytes, whole) walks with a cursor; the rewrite path asks
for the whole file, the read paths keep reading as before.
- Python: push_all already reads the whole file (unpackb refuses trailing
bytes, the legacy decoder refuses anything but a list of maps); pinned by a
test.
Tests in each runtime run the same file matrix (truncated, trailing, scalar,
map, string first; empty, empty array, legacy), plus a C# unconditional commit
over a store with trailing bytes and the C# Mesh raw write.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
SingleFileBackingStore.PushAll replaced the file without reading it, so a
truncated store or one in a format this runtime does not read was clobbered,
where Rust, TypeScript and Python refuse. PushAll now reads the header of a
non-empty file under the lock before it writes; a file it cannot read throws
and is left as it was. An empty file and a legacy envelope array carry no
header and are still replaced.
The header is read by the store's codec: SingleFileBackingStore gains a
protected virtual ReadStoreHeader (default: read the whole snapshot), and the
MessagePack store overrides it with
CultDocumentMessagePackSerialization.ReadStoreHeader. An unconditional commit
reads the file it rewrites through the same check.
Doc text: last-writer-wins now says it holds only over a file the writer can
read (CultCache.cs PushAll comment, cultcache-persistence-format.md
"Conditional commit is the only safe multi-process write" and writer rule 4,
cultcache-store-composition.md "A plain flush and an unconditional commit are
last-writer-wins").
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A record's HoldsIds was decided by its admission and trusted afterwards, so a
document changed after it was committed, or loaded and changed in place,
flushed under the header its admission-time content needed (Soul P7, P10).
HoldsIds now has one owner, the record's stored form, read when asked:
- a plain record: CultElementIds.Holds over its document, a scan that reads
and refuses nothing;
- a variant: the same scan over each override value, decoded with the codec
the record was built with (the cache's, or the attached cache's at load);
- a variant built with no codec (a store that loaded it with no cache) cannot
see its values and counts as holding an id, so a store never writes an
unmarked header over ids it could not see.
Cut: the admission path that set HoldsIds on plain records, on variants from
their resolved copies, and on flattened records; the internal setter and its
backing flag; IdPlan.HeldIds and IdPlan.HoldsIds. The Mesh typed write reads
Holds off the document it writes. Admission keeps its own id counting
(IdPlan.Count) for IdsInMemoryOnly.
Tests: an id added to a committed or loaded deck after admission marks a
PushAll v3, and taking it away unmarks it; a variant read without a codec
counts as holding ids while the same record read with one does not.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
One conflict, at the end of Admit in CultCache.cs. The branch had added the
IdsInMemoryOnly marking of a load's minted ids before Apply; main had turned
the thread's hold into a Hold carrying Changes and Failures and run the
journals right after Apply. Both are kept, in that order: the marking touches
only the records and runs no user code, then Apply mints the Sequences, the
hold records the changes, and the journals run under the gate.
Id planning and Project do not collide. Ids are planned and applied first,
Resolve (variants, then Project, then the name and index rules) reads the
filled documents, land runs, and only then Apply mints. A Project getter that
throws is caught by the same handler that undoes the id plans, so a refused
admission changes no document and mints nothing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
rmp_serde refuses a sequence visitor that stops early, so the prefix-only header read the
previous commit introduced could not decode any store. store_header is now one visitor: the
first slot is the header (None when it is not a string, the legacy envelope array), the rest
is walked, and a truncated file is an error.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ExistingHeader read the whole snapshot and swallowed every failure into "no header": a
truncated v3 file got a v1 raw write and an unknown v9 file was overwritten, where
TypeScript, Python and Rust refuse. CultDocumentMessagePackSerialization.ReadStoreHeader
reads the prefix header, refuses a file that is not one complete MessagePack value or whose
format this runtime cannot read, and the raw write keeps a whole marked file marked.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A CultStoredDocument built by a caller and handed straight to a store never passed an
admission, so HoldsIds stayed false and its ids landed under a v1/v4 header. An admission
still decides it (it counts the ids it fills); a record that never had one reads it off the
document it stores. CacheBackingStore.HeaderFor documents the rule for third-party stores.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
store_header returned None for any bytes it could not walk, so push_all read a truncated
variant store as legacy and overwrote it v1. It now proves the file is one complete
MessagePack value first (an error otherwise, the rewrite refuses and leaves the file), then
reads the header from the prefix; the reader's own header read no longer walks to the end.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The behaviour was already a refusal that leaves the file untouched; now a test fails if a rewrite
ever reads the disk header less strictly than a load does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
push_all reads the file and reuses store_header and readable_store_format, the same pair a load
uses. The 96-byte prefix parser it replaces misread a str16 header and any str8 header past the
prefix as v1 and overwrote a store it could not read. Pins both encodings refused with the file
untouched.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CacheBackingStore.HeaderFor(holdsIds, existingHeader, wholeStore, directoryStore) replaces FormatFor,
DirectoryFormatFor and the writers' private marker logic. A writer that rewrites the whole store
decides by content; a writer that touches part of it keeps a header already marked. Flush,
unconditional and conditional commit, the directory store and CultMesh's single-file write all call
it. A typed CultMesh write replaces the file with a document it can see, so it decides by content;
a raw payload is opaque and keeps a marked file marked.
A record's HoldsIds is what its stored form persists: a load keeps a variant's delta as read, so
the ids a load mints in an override exist only in memory and no longer mark the store. A
conditional commit onto the file no longer reads this cache's view of what the file held, so a
stale view cannot over-mark it. IdPlan.HoldsIds is derived from HeldIds and the fills.
Reverts InternalsVisibleTo GameCult.Caching.MessagePack and the public DirectoryFormatV4/V5
constants: the directory store reaches the decision through the protected member it inherits.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
push_all overwrote a v9, variant-v2 or variant store with a v1 file, dropping its variants; TypeScript and Python
refuse. The rewrite header now comes from the file's own prefix and shares one readable-format rule with the
decoder. TypeScript pullAll resets its header when the file is gone, empty or legacy, as Python does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A flattened variant's HoldsIds is read off the document re-resolved against the post-batch base, not carried from the
variant or from the document the flatten was planned against. The conditional commit onto a file decides by the same
rule as the unconditional commit (ids in the records the file will hold, ids only this cache minted excepted, a marked
file staying marked). The directory manifest's header choice moves beside FormatFor. Pins the three-batch flatten
probe on the commit and flush paths, the conditional commit, the last id leaving, an id-bearing override that is not
the last, and a failed CultMesh write giving its ids back.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
TS pushAll wrote the header its instance last read (v1 when fresh) without reading the
file. It now reads the on-disk header first, as push and delete do through pullAll. Python
resets its remembered header when the file is gone or empty, so the disk decides.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A store is written v3 (v5 for a directory) only when a record in it holds a non-empty
element id. CultElementIds.Plan reports that as IdPlan.HoldsIds; the admission stamps it
on the stored record and every store reads it. The descriptor's CarriesElementIds is
deleted. A file already marked on disk stays marked. CultPersistedStoreSnapshot.FormatFor
is the one header decision.
Plan refuses one element object sitting twice in a list at write, naming the record and
list, instead of writing both ids and refusing the store at load.
CultMesh single-file write mints, refuses and marks as a cache write does; the read applies
the load-time plan, so a write and a load refuse the same things.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
old-reader-refusal.sh stages the C0 merge (e382bb4, or OLD_READER_REV) plus the v3 vector
and one probe per runtime onto a commit; the probe run from that tree is the old reader.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
A store that can hold element ids declares cultcache.store.v3. These runtimes pass payload
bytes through, so they read it and write back the header they found; they do not decide
when a document carries ids. Vector: tests/vectors/document-variants-c2a/v3-base.msgpack.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Engine tests for the plan (duplicates, format, empty derived source, atomicity, two
cultures, interface unions, key sharing); cache tests for the marker on single-file and
directory stores, override values on write and load, minting that is given back, and the
in-memory-only mark.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The id rule is one plan, made before anything is written: Plan decides every fill and
refuses a duplicate, a random id that is not 12 lowercase hex characters, and a derived
id whose source is null or empty, naming the record and list (CultElementIdException).
A load refuses what a write refuses. Apply and Undo make minting part of the admission:
a refused or uncommitted write leaves the caller's documents as they were.
Override values mint inside Resolve, into the decoded copy, so the public
CultVariantOverride.Set, C1-era overrides and Override() all get ids on write and on load;
Override() no longer mints. Interface unions are judged on their concrete types, the same
rule registration and minting share, and a subtype reusing the id's key is refused at
registration. Dictionary keys format with the invariant culture.
Records minted at load are marked per entry (IdsInMemoryOnly) instead of listed by key, so
a deleted record leaves the mark with it, and a flush of the whole store clears it. A store
whose types reach an object list declares cultcache.store.v3 (directory: v5), which every
reader older than ids refuses; a marked file stays marked when a writer merges onto it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The C# cultcache interop peer gains an object-list document (a deck of marks).
C# writes it with cache-minted ids; TypeScript, Python and Rust read the ids as
ordinary members, and TypeScript-written ids survive a C# read.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
CultWitnessContractPin, CultWitnessArtifactEntry and CultWitnessTimingEntry get
an id slot. The JSON schema contract and the Python schema copy gain an optional
id per element, and the Python and TypeScript fixtures carry and assert ids.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
[CultElementId(nameof(Member))] declares an id derived from that member's
invariant text, never random, so content-addressed elements keep encoding to the
same bytes. CultMeshCdnChunkRef derives its id from Offset; PackArtifact sets
it, the Rust pack_content writes it, and the committed content vectors are
re-cut once so both runtimes carry it. Engine tests drive Problems and Assign
directly (InternalsVisibleTo the Caching tests).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
[CultElementId] marks a string member; the registry refuses a document type
whose list element type (union bases included) lacks one, naming the type and
member. A write mints a random 12-hex id for each unset id and refuses a
duplicate within one list. A store written before ids loads with ids minted in
memory from SHA-256(record key, element path), so every reader and every reload
agrees; they persist on the store's first write. CultCache.MintElementIds()
rewrites the records whose ids exist only in memory.
CultLib's own list-element types gain an id (Geometry, Mesh body, discovery,
entity prefab). CultMeshCdnChunkRef and the CultWitness* elements are pinned by
cross-runtime vectors and are left alone; see the C2a report.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-09-29 12:31:12 +02:00
483 changed files with 55921 additions and 3962 deletions
`unowned_schema` (a write, or a remote put or delete, for a schema and key no shard owns),
`write_not_authorized` (a peer's put or delete its server's remote write authority did not grant) and
`unsupported_schema_version` (a message whose `schemaVersion` the peer does not speak; `details` names `schemaVersion` and the offered value) and
`limit_reached` (a receipt-capable server's per-session remembered-write table is full, so a new write is refused; `details` all null) and
`message_invalid` (a request that lacks a required field or carries an ill-typed one: a document-put or delete, or a subscription without an id; `details.field` names the request field, or `message` when the whole frame cannot be read).
A malformed put or delete is `message_invalid`, not a per-operation code: the peer knows which operation it sent.
## Live Invariants
@ -122,6 +127,9 @@ copies), the C# `CultNetErrorMessage` factory, the Rust `CultNetErrorCode`, and
- One primary owns writes for a shard until a stronger policy exists.
- Non-owner writes are rejected or explicitly forwarded. They are not silently
applied.
- A peer's put or delete is applied or forwarded only when the server's
`RemoteWriteAuthority` grants it; none configured refuses every remote write.
The connection key admits a peer to the transport and is never a grant.
- Raw snapshot/put/delete messages must pass through `CultNetDatabase`, not
bypass shard policy.
- The accepted shard log is stored in replica catch-up wire form. Object logs
"description":"Machine-readable refusal code (R-N). Additive: a v0 peer that does not know this field ignores it. The set is a closed enum: Rust and TypeScript refuse a message whose code is not listed, so adding a code is one coordinated change across every runtime (this schema, the C# factory, the Rust CultNetErrorCode, and a captured-bytes case for each).",
"description":"Code-specific structured detail (R-N). Every field is populated only for the code that defines it; the others are null, not omitted.",
@ -36,8 +36,8 @@
"additionalProperties":false,
"required":["field","value","asOf","current"],
"properties":{
"field":{"description":"selection_invalid: the selection field that failed. variant_unsupported: \"subscriptionId\", when a subscription was ended. unowned_schema: \"schemaId\".","type":["string","null"]},
"value":{"description":"selection_invalid: the offending value, when there is one string worth naming. variant_unsupported: the id of the subscription that ended. unowned_schema: the schema id no shard owns.","type":["string","null"]},
"field":{"description":"selection_invalid: the selection field that failed. variant_unsupported: \"subscriptionId\", when a subscription was ended. unowned_schema: \"schemaId\". write_not_authorized: \"recordKey\". unsupported_schema_version: \"schemaVersion\". message_invalid: the request field that is missing or ill-typed (\"message\" when the whole frame cannot be read).","type":["string","null"]},
"value":{"description":"selection_invalid: the offending value, when there is one string worth naming. variant_unsupported: the id of the subscription that ended. unowned_schema: the schema id no shard owns. write_not_authorized: the key of the refused write. unsupported_schema_version: the schemaVersion the peer does not speak.","type":["string","null"]},
"asOf":{"description":"cursor_stale: the cursor's own minted asOf.","type":["integer","null"],"minimum":0},
"current":{"description":"cursor_stale: the answering server's current asOf.","type":["integer","null"],"minimum":0}
"description":"cultnet.error.v0 plus messageId (write-receipt.md). Sent in place of cultnet.error.v0 only to a connection that declared the write-receipt capability. v0 is unchanged: its schema is closed (additionalProperties false), so an extra key would make a v0 peer reject the frame instead of ignoring it. Map-mode: every key is written, messageId nil when the error answers no readable message. The code enum is the v0 enum plus limit_reached (write-receipt.md, Remembered-write limits).",
"type":"object",
"required":[
"schemaVersion",
"error",
"routingHint",
"code",
"details",
"messageId"
],
"additionalProperties":false,
"properties":{
"schemaVersion":{
"const":"cultnet.error.v1"
},
"error":{
"type":"string",
"minLength":1
},
"routingHint":{
"description":"Optional shard routing detail for an authority failure. Null for every selection/cursor refusal code (R-N).",
"type":[
"object",
"null"
],
"additionalProperties":false,
"required":[
"shard",
"reason"
],
"properties":{
"shard":{
"type":[
"object",
"null"
]
},
"reason":{
"type":"string"
}
}
},
"code":{
"description":"Machine-readable refusal code (R-N). Additive: a v0 peer that does not know this field ignores it. The set is a closed enum: Rust and TypeScript refuse a message whose code is not listed, so adding a code is one coordinated change across every runtime (this schema, the C# factory, the Rust CultNetErrorCode, and a captured-bytes case for each). limit_reached: a receipt-capable server refuses a NEW write because its per-session remembered-write table (1024 entries) or its server-wide memory budget is full; messageId is the write's id.",
"type":[
"string",
"null"
],
"enum":[
null,
"selection_invalid",
"cursor_stale",
"cursor_invalid",
"reference_outside_target",
"variant_unsupported",
"unowned_schema",
"write_not_authorized",
"unsupported_schema_version",
"message_invalid",
"limit_reached"
]
},
"details":{
"description":"Code-specific structured detail (R-N). Every field is populated only for the code that defines it; the others are null, not omitted.",
"type":[
"object",
"null"
],
"additionalProperties":false,
"required":[
"field",
"value",
"asOf",
"current"
],
"properties":{
"field":{
"description":"selection_invalid: the selection field that failed. variant_unsupported: \"subscriptionId\", when a subscription was ended. unowned_schema: \"schemaId\". write_not_authorized: \"recordKey\". unsupported_schema_version: \"schemaVersion\". message_invalid: the request field that is missing or ill-typed (\"message\" when the whole frame cannot be read). limit_reached: no field; details stay all null. message_invalid also names \"messageId\" (over 256 bytes), \"document\", \"schemaId\" or \"channel\" (write on an unreliable channel).",
"type":[
"string",
"null"
]
},
"value":{
"description":"selection_invalid: the offending value, when there is one string worth naming. variant_unsupported: the id of the subscription that ended. unowned_schema: the schema id no shard owns. write_not_authorized: the key of the refused write. unsupported_schema_version: the schemaVersion the peer does not speak.",
"type":[
"string",
"null"
]
},
"asOf":{
"description":"cursor_stale: the cursor's own minted asOf.",
"type":[
"integer",
"null"
],
"minimum":0
},
"current":{
"description":"cursor_stale: the answering server's current asOf.",
"type":[
"integer",
"null"
],
"minimum":0
}
}
},
"messageId":{
"description":"The messageId of the write (or other request) this error answers, echoed byte for byte. Null only when the frame was unreadable or carried no messageId.",
"description":"The receipt for one admitted document put or delete (write-receipt.md). Sent only after the write's commit point, only to a connection that declared the write-receipt capability, once per messageId. Map-mode: every key is written, none is nil. A receipt never carries an error, and an error.v1 never carries admission.",
"provenance":"Captured from MessagePack C# 3.1.7 (CultNetErrorMessage/CultNetErrorDetails, Options = Standard.WithSecurity(UntrustedData)); pinned byte-for-byte by packages/cultnet-rs/tests/error_message.rs, whose inline hex equals messagePackHex here (test error_vectors_file_matches_the_inline_csharp_captures). Rust encodes to these exact bytes, so C#-written and Rust-written bytes are one set. Every key is written on the wire, nil included.",
Status: specification. No runtime implements it yet; the conformance follow-ups below do.
Rules: Heimdall's ruling of 2026-11-08 (a write completes on its result frame; an ack or a return is not admission).
Files here: `cultnet.write-admitted.schema.json`, `cultnet.error.v1.schema.json`, `write-receipt-vectors.json`. They move to `contracts/cultnet/` when the first runtime implements them.
## The problem
A put or delete today has two answers on the wire: nothing (admitted, or the server never read it) and `cultnet.error.v0` (refused, with no id saying which write). A transport ack proves delivery. Clients that completed on the ack, or on a window with no refusal, reported refused writes as done (CultLib #55 to #58 bound this with a result window; this spec replaces the window with an answer).
## Frames
**`cultnet.write_admitted.v0`**: `{ schemaVersion, messageId, operation: "put"|"delete", schemaId, recordKey }`, every key written, none nil. Sent once a put or delete is admitted. The id echoes the write's `messageId` byte for byte; `operation`, `schemaId`, `recordKey` echo the write.
**`cultnet.error.v1`**: `cultnet.error.v0` plus a required `messageId` (string or nil). The code enum is v0's plus `limit_reached` (see Remembered-write limits); `cultnet.error.v1.schema.json` is the authority for the closed set. Sent in place of `error.v0` to a receipt-capable connection. `messageId` is the id of the request the error answers; nil only when the frame was unreadable or carried no id.
Why a new error version and not an extra key on v0: the v0 schema is closed (`additionalProperties: false`) and the TypeScript decoder (Ajv over the closed schemas) rejects unknown keys and unknown `schemaVersion`s, and `parseCultNetMessage` throws on an unknown version; Rust's decoder errors on an unknown version too. Unknown-key tolerance of the Rust, Python and Kotlin error decoders was not checked, so none is assumed, so a surprise key or frame breaks an old client's whole receive path. Old clients are therefore never sent either new frame. This reverses the "old clients ignore it" assumption, which is false for today's decoders.
## Capability: opting in
A connection is receipt-capable when its `cultnet.hello.v0` (sent before its first write; the schema channel is reliable and ordered) lists both `cultnet.write_admitted.v0` and `cultnet.error.v1` in `supportedMessageVersions`. A server MUST NOT send either new frame to any other connection, and answers it exactly as before (no receipt, `error.v0`). A receipt-capable server lists both in its own hello, so a client can tell a legacy server from a receipt server that has not answered yet. A client that sends no hello gets legacy behaviour. A server advertises the two frames in its hello only if it implements the receipt rules end to end.
**The capability changes framing only; it never authorizes.** A hello authenticates nothing and grants nothing: who may write, and which keys, is decided by the server from the transport-authenticated actor exactly as without the capability. Declaring the capability never makes a refused write admitted, never widens an actor's authority, and never changes what a write does; it selects only which frame the answer arrives in.
## Server rules
1. **One answer.** For every put or delete on a receipt-capable connection the server sends exactly one of `write_admitted` or `error.v1`, carrying that write's `messageId`. Never both, never two different answers.
2. **After the commit point.**`write_admitted` is sent only after the write is committed by the same rule that makes it visible to readers. Receipt of the bytes, parse, or authorisation alone is not admission.
3. **Id on every refusal.** Every refusal of a write, whatever its code (`write_not_authorized`, `message_invalid`, `unowned_schema`, `unsupported_schema_version`, shard authority errors, ...), is an `error.v1` whose `messageId` is the write's id. When the frame is readable enough to find the id, it is found; `message_invalid` for a missing or ill-typed field still carries the id. Nil only when no id can be read.
4. **Batches.** There is no batch frame; a batch is N writes, each with its own `messageId`, each answered under rules 1 to 3. A failed batch names its member by that id. A server may not collapse a batch into one answer.
5. **Identity and idempotent resend.** A `messageId` is unique per authenticated actor and session: the client mints it, never reuses it for different content in that session, and a server never accepts the id of one actor as the id of another. The server's dedupe key is `(actor, messageId)`, where *actor* is the transport-authenticated identity, never a name the hello or the frame claims. Alongside the key it keeps the **content hash** of the write and the answer it gave. The content hash is SHA-256 of the MessagePack array `[operation, schemaId, recordKey, payload]`, where `payload` is the raw document payload as a MessagePack `bin` for a put and nil for a delete (the vectors give worked hashes). **Scope: at least the session.** The key lives for the life of the session it was made in, and a server may bound how many writes it remembers per session (see Remembered-write limits; a server never forgets an id it promised to answer). A server MAY keep the key longer for the same actor, up to `DEDUPE_RETENTION_MS` (10 minutes), but a client MUST NOT rely on it: **a reconnect is a new session, and a resend on the new session is a new write.**
- Same key, same content hash, answer was `write_admitted`: answer `write_admitted` again with the same body. The write is not applied twice.
- Same key, same content hash, answer was a refusal: the same refusal again.
- **Same key, different content hash: refuse `message_invalid` (details `field: "messageId"`) and never re-send the old receipt.** Old receipts are for the old content; showing one for new content would report an unapplied write as admitted. The original write's state is unchanged.
- **Id size.** A write's `messageId` is at most 256 bytes of UTF-8. A longer id is refused `message_invalid` (`details.field: "messageId"`, the refusal's `messageId` the id as received, text fixed per Fixed texts) BEFORE anything is remembered: no table entry, no digest, no budget charge. Such a refusal is therefore not replayed from the table; a resend is refused afresh, identically.
- **What is remembered.** Per entry a server keeps only fixed-size digests (SHA-256) of the `messageId` and of the content hash, plus the answer (`admitted`, or the refusal's code and details). Memory per entry is bounded independent of id length and payload size. Lookup compares digests.
- **Remembered-write limits.** The per-session table holds at most 1024 entries. A server MAY also enforce a server-wide memory budget across all sessions (reference: 32 MiB of remembered entries). A NEW write (a key not already in the table) that would exceed either bound is refused `limit_reached` for that write alone, `messageId` the write's id, `details` all null; the write is not applied and the session continues. A resend of a remembered key is still answered from the table.
- **Limit refusals are remembered.** A `limit_reached` refusal is remembered like any other: a resend of the same id and content hash replays `limit_reached`. No slot is freed mid-session, so an id refused at a bound never becomes admitted later in that session (a server at the cap keeps the refused digest pair in a bounded refusal set of at most 1024 per session, outside the entry budget; past it, a further new write is refused `limit_reached` without being remembered, and the session refuses every id it has not seen for the rest of its life, even if a server-wide budget has since freed, so a refusal that was not remembered is never followed by an admission of the same id). The client mints a new id on a new session.
- **Required fields.** A put or delete whose document or `schemaId` is null or missing is refused `message_invalid` with `details.field``"document"` or `"schemaId"` (the first missing in that order) and fixed text; the write is not applied. A `schemaId` or `recordKey` that is empty or only whitespace (present but names nothing) is not "missing": the write is refused `message_invalid` exactly as a frame that does not read, with `details.field``"message"` and the write's id (vector `error_v1_empty_schema_id_is_unreadable_and_names_the_write`), and is never applied; a server must not hand it to its store.
- **Fixed texts.** The `error` string of these refusals is fixed, so vectors can compare it: `message_invalid: messageId exceeds 256 bytes`, `message_invalid: document is required`, `message_invalid: schemaId is required`, `message_invalid: receipt sessions require the reliable channel`, `limit_reached: the session's remembered-write limit is full`.
- **Reliable channel only.** A write is accepted on a receipt session only on a channel that is reliable and not sequenced (reliable ordered and reliable unordered both qualify; a sequenced channel drops superseded packets, so a write on it can vanish without trace). A write arriving on any other channel is refused `message_invalid` with `details.field: "channel"` (the transport profile's word for it) and fixed text. Receipts and refusals travel the reliable channel; a write that could be lost without trace cannot be answered once.
- Past the session (or a server's longer retention, if it keeps one) a resend is a new write.
6. **Forwarded writes.** A server that forwards a non-primary write answers the original peer only with what the primary answered: the primary's receipt becomes `write_admitted` with the original id, its refusal becomes `error.v1` with the original id. If the forward gets no answer the server relays nothing: no receipt, no error, no synthesized refusal. The client's own rule below then reports the write as unconfirmed. A forwarder never turns an absent or ambiguous primary answer into an admission.
7. **Looks alike.** An admitted and a refused write differ in `schemaVersion`; a receipt has no `error`, an `error.v1` has no admission. A transport ack is neither.
## Client rules: three outcomes
A write settles as exactly one of:
- **admitted**: a `write_admitted` whose `messageId` is the write's.
- **refused**: an `error.v1` whose `messageId` is the write's, carrying its `code` and `details` as the typed error.
- **unconfirmed**: nothing settled it. Reasons: `window_elapsed` (the result window ran from the *send* and passed with no answer; see Result window), `session_ended` (the peer closed the session, or it ended, before the write was answered) and `local_send_failed` (the client's own send of this write failed). A local send failure is never reported as refused: no server answer exists, and the write may or may not have left the machine. Unconfirmed is not success and not refusal. APIs that return a value return the tri-state; APIs that throw throw a distinct refused error and a distinct unconfirmed error and return normally only for admitted. A caller that needs admission treats unconfirmed as not admitted. Resend with the same `messageId` and unchanged content, **on the same session**, is safe against a receipt-capable server (rule 5). After a reconnect the resend is a new write, and a legacy server may apply a resend twice: the caller resends only writes that are idempotent by record key (a put of the same content to the same key, a delete of a key), and reads the record back before resending any other.
**Result window.** Two named constants (also in the vectors): `RECEIPT_RESULT_WINDOW_MS = 5000`, the default against a receipt-capable server (its hello lists both frames); `LEGACY_RESULT_WINDOW_MS = 500`, used only as the fallback against a server whose hello does not list them (or that sent none). The window runs from the send. A receipt-capable server that has not answered within its window is reported as unconfirmed like any other; clients log it as a server that missed its window.
**Late answers.** An answer for a write that already settled `unconfirmed` does not change what the caller was told, but the client records it on the write: a late `write_admitted` makes the write's recorded state `admitted_late`, a late refusal makes it `refused_late`. The write handle exposes that state, and a caller that got `unconfirmed` MUST consult it (and the record) before resending: it must not blind-resend a write that was admitted late. A resend on the same session with the same id and content is harmless on a receipt-capable server (rule 5); across a reconnect the caller follows the idempotence rule above.
**Legacy answers.** An `error.v0`, or an `error.v1` with nil `messageId`, names no write. It is attributed to a write only when exactly one write on the connection is unanswered; it then refuses that write. With two or more unanswered writes it settles none of them and each ends unconfirmed at its window. A frame naming a `messageId` that is not in flight settles nothing.
**Batches can commit partially.** A batch is N independent writes with no atomicity: some members may be admitted while others are refused or unanswered, and nothing is rolled back. The client reports each member's outcome and a batch result is only their set: admitted if every member is admitted, refused if any member is refused (naming the members), otherwise unconfirmed (naming the unanswered members). The caller owns retry: it resends exactly the members that are not admitted, with their original ids.
## Changes to what #55 to #58 shipped
Those clients complete *successfully* when the window ends with no refusal. Under this spec that case is `unconfirmed`. The TypeScript one-shot writer (#62) already reports it so: it resolves `admitted` only on a `write_admitted` carrying the put's id, and no other frame, a put echoed under the same id included, ever counts as admission. A caller whose server cannot answer a write yet (Odin, until it is receipt-capable) opts into the explicit send-only mode per call (`sendOnly: true`), which resolves the distinct outcome `sent` once the transport has acknowledged the put and never `admitted`. The Idunn presence publisher is send-only by its own function (`sendPresenceDocument`, not the `sendOnly` option): the same `sent` result, plus a rejection if a refusal naming the put (or naming nothing) arrives by the transport ack. Receipts and send-only are both opt-in, neither is the default. The C# and Rust clients of #55 to #58 keep the window-success until their conformance work lands, and callers of those must not read it as admission.
## Vectors
`write-receipt-vectors.json`: `constants` (the named windows and retention), `contentHash` (worked dedupe hashes), `dedupe` (server replay cases for rule 5), `wire` (MessagePack hex plus the decoded message for each new frame; the bytes are spec-authored with msgpack-python and every runtime's first conformance PR re-captures them with its own encoder and must match), `schemaInvalid` (frames that must be refused by a validator, each with why), and `scenarios` (event replays with the outcome each write must settle as; the clock is abstract; each case names its window constant). A conformant client passes every scenario; a conformant server's answers for the same inputs satisfy server rules 1 to 7.
## Conformance status
| Server class | Advertises both frames | Server-wide memory budget (MAY, Remembered-write limits) |
|---|---|---|
| `CultNetDatabaseServer` | yes | not implemented: only the per-session limit of 1024 applies |
| Shard write forwarder | yes | not implemented: only the per-session limit of 1024 applies |
| C# RUDP schema server (`CultNetRudpSchemaServer`, `src/GameCult.Networking/CultNetRudpSchemaServer.cs`) | no | n/a |
Advertising is per server class, not per transport. A class advertises both frames in its hello only if it implements the receipt rules end to end; the C# RUDP schema server stays "no" until it answers writes with receipts.
## Conformance follow-ups (not in this PR)
C# server and client, Rust, TypeScript, Python, Kotlin: emit and parse the two frames; hello capability; tri-state client outcome; replace the #55 to #58 window-success; re-capture the wire vectors in each runtime. Server write paths include the shard write forwarder (rule 6) and the Odin catalog.
Brokkr, Aquarium, Ymir, AquaSynth and Mimir have no `CultLibRevision`/`VerifyCultLibRevision` guard,
and each builds against the sibling `..\CultLib` checkout. In its own repo, each adopts Aetheria's
guard at the 2.0.0 tag, with one Hands per repo, before the merge reaches main. The pins touch no
CultLib file, so they can start now and run in parallel with steps 1-3. Eve and Gjallar get no pin
(ruled). The step is done when each repo's guard fails a build against a CultLib checkout at
another revision and passes at the tag, and the follow-up is resolved.
5. **Main into the branch, then full verification.** After steps 1, 2 and 4, `git merge origin/main`
into `hands/variants-c2a`. It should be clean, because main's commits since 02cf04b5 are docs and
research only. Re-check that before merging (`git diff --stat 02cf04b5 origin/main -- src tests
packages unity`). Then run every suite at the merged head on Yggdrasil: Caching, Mesh, Networking,
Geometry, WebSockets, ConsumerBuild, InteropPeer, cultcache-ts/-rs/-py, cultnet-ts and cultmesh-ts.
Run `old-reader-refusal.sh gate` for csharp, typescript, python and rust. If main gained code by then,
a Soul pass covers the merge.
6. **Land.** The land is `git merge --no-ff`, never a squash, so the SHAs that every report and verdict cites stay reachable from main (skip the branch's non-building commits with `git bisect --first-parent`). Steps 1 and 2 closed at 723b7996 (claimant-one-owner, land-c2a-gate); step 5 closed at the merge commit 3157d4b3 (Soul `cut-land-c2a-main.s1` holds). Step 3 closed except `cut-land-c2a.s1.stale-unity-plugin-dlls` and `cut-land-c2a-gate.s2.mutation-table-reads-gate-as-zero`, which wait on the 3.0.0 release cut and the Eureka tools respectively; step 4 closed with `consumer-pins-before-land` (pins since 2026-10-22). Run `git merge --no-ff hands/variants-c2a` onto main, with an empty diff against the
verified head. There is no tag: tags, the Unity plugin rebuild and `declared-range-gate-own-version`
belong to the 3.0.0 release cut. Open follow-ups `c2a-load-slower` (load 2 to 2.3 times slower than
main on a synthetic Aetheria catalog) and `drift-write-ownership-leaves` do not say "before land".
`c2a-load-slower` still needs a reason or a fix before the 3.0.0 release, under
`IReadOnlyList<EveCommandTemplate>`, 0 ids in the repo), so C2a registration refuses its
documents as soon as the checkout pulls the land. Grep finds no object lists in Brokkr's
`[CultDocument]` files (not build-verified). Removed C2a API (`Options`, `AsSchemaAlias`) is
unused by Eve and Brokkr. Fork: `variants:question:unpinned-consumers-at-land`.
## Classification
| Item | Class | Why |
|---|---|---|
| c2a-format-version-gate | **a** | The gate test above. No new header: v3/v5 already refuse. |
| slot-reuse-contract-doc | **a** | The ruled consequence of the write set C2a lands. One paragraph in cultcache-schema-compatibility.md, absent at a5c7954e (grep). |
| foreign-ids-under-v1-header | **b** | Obsoleted by fb065f0d (write-time `CopiedHoldIds`, after the finding) for every C# writer with the type. The residual (non-C# or opaque writers) is what `non-csharp-writers-mark-ids-2` ruled to carry, and the contract states it (persistence-format.md:434-439). The gate changes neither side, so no fork remains. Resolve as Superseded by that ruling. |
| custom-resolver-element-ids | **b** | Fixed by b091d21f (19:52Z, after the 18:27Z follow-up): `ListElementsClaimedByAResolverAreRefused`, `AResolverClaimingTheListTypeOfAnObjectListIsRefused`. s6 holds on the claim loop. |
| consumer-element-id-sweep-r2 | **c** | Each repo's sweep is its own work (q3). Only the unpinned trackers matter at land, and that is the question below. Aetheria bumps a pin on purpose (C4). |
| schema-discovery-holder-recogniser | **c** | `TryGetDictionaryRefTypes` is identical on main (CultCache.cs:1288). It describes record-ref dictionaries and is not on the id path. Changing it moves schema hashes, which is its own epoch event. |
| copy-check-memo | **c** | Performance only (about 3x flush on unmarked stores of 20k records). Correct as is. |
| string-key-field-and-shadow-pins | **c** | No wrong behaviour at 64ec14f9. These are test pins and may ride the land cut if Hands touches the file. |
| declared-range-gate-own-version, range-check-in-ci | **c** | The script exists only on C2a and runs in no CI. It guards a TS publish, not a merge. Fix both before the first TS release from main after the land. |
| s6 Unproven: Unity/IL2CPP identity, other MessagePack versions | **c** | The merge reaches no Unity build. Unity sees C2a only at the 3.0.0 tag, so measure it in the Unity release cut (operator Unity check, C3). |
| multi-store-commit-atomicity, store-replace-cut, reader-refuses-duplicate-keys, soft-flush-all-runtimes, writer-dangling-link-tolerance, unique-index-cut, py-shared-write-lock, rust-lock-open-waits, ts-lock-test-hardening, c0/legacy/directory vectors, culture-test, wire-id-cuts, remaining-variant-cuts, golden-path, paged store | **c** | None says "before land". Each is pre-existing on main or C2a-independent. None lets a released reader strip ids or corrupt a C2a store. |
## The land cut's shape (transcribe once the question is ruled)
- **Base.** Merge main at or after the `cultlib-unity-v2.0.0` tag into `hands/variants-c2a`.
Versions stay main's. CHANGELOGs get Unreleased entries, and the Unity entry says major.
- **Deletes and collapses.** Delete `PersistedSchemaId`, `PersistedIdentity`,
`CultStoredDocument.MarkWritten` and `CacheBackingStore.MarkWritten` (main, about 60 lines),
and the store-composition contract's MarkWritten paragraph. C2a's `StoredVersion`/`Wrote`
is the one owner. In TS, delete C2a's header handling in
single-file-messagepack-backing-store.ts and move it into store-file.ts. In Rust, C2a's v3
decode goes through the grown decoder.
- **Adds.** Retarget old-reader-refusal.sh to the tag. Add the C# probe's never-rewritten
assertion and the v5 directory case (as in gate1.log). Add the slot-reuse paragraph.
Port main's ConditionalCommitTests to the version identity.
- **Verification.** Full suites: Caching, Mesh, Networking, Geometry, WebSockets,
ConsumerBuild, InteropPeer, cultcache-ts/-rs/-py, cultnet-ts and cultmesh-ts, all on
Yggdrasil. Run old-reader-refusal.sh for all four runtimes at the tag, and it must refuse.
Negative greps over the merged tree: `PersistedSchemaId|MarkWritten`, conflict markers.
Mutation: drop the v3/v5 stamp in `HeaderFor` and the gate probe must fail.
- **Estimate.** About 950 conflict lines resolved. Net -60 to -150 from the identity
collapse and TS de-duplication, +80 to +120 tests and docs. One Hands pass of 100-150
calls, then one Soul pass.
- **Precondition (question).** Eve.Surface, and Brokkr if confirmed, are pinned or swept
before the merge reaches main.
## Store gate: attachment is the attach's hold (store-gate-one-owner r2)
Yggdrasil `probe-base.log`, 2026-10-25 around 23:00 CEST. "Disagree" means that after the
attach, a commit of `later` and a flush, the attaching cache holds `a,later` and a fresh cache
over the same file reads something else.
| Entry point (line) | Fact it reads | In the attach window at 0fa836cb |
| --- | --- | --- |
| Single-file `PullAll` (4156) | `Gate`: `Cache` guessed unlocked, then re-read under the re-entered store gate | Runs as detached; its `Loaded` re-enters the cache and is refused as a journal write. Agrees. |
| Single-file `Push` (4212) | same | Runs as detached and stages `b`. Disagree: fresh reads `a,b,later`. |
| Single-file `Delete` (4223) | same | Runs as detached and stages a removal the cache never made (dirty after the attach). |
| Single-file `PushAll` (4239), and `Dispose` (3722) through it | same, then in-body `Cache == null` (4531, 4543, 4588, 4603, 4611, 4661) | Runs as a cacheless writer. Push then PushAll writes `b` to the file inside the attach (disagree). Delete then PushAll is refused with the cacheless writer's text. On a variant store it is refused as "a writer with no CultCache" (Soul s4). |
| Single-file `CommitBatch` (4256) | same | Runs as detached and commits `b`. Disagree. |
| Single-file `JudgeChecks` (4568), `RefuseRemovalOfUndeclared` and `StageChecks` (4558, 3915) | same | Run as detached. Agree in the probe, but under the wrong gate's rules. |
| Single-file `ReadStoreSnapshot` (4687) | same | Runs as detached and reads the file. |
| `WriteRecord` (4342), `WriteDocument` (4366) | `InStoreHold`'s per-thread hold (cacheless re-entry), before `Cache` | Refused with the re-entry text (h5). Agree. |
| Directory `PullAll` (48) | `Gate`, as above | Refused as a journal write. Agree. |
| Directory `Push` (147), `Delete` (160), `CommitBatch` (172) | same | Run as detached. Disagree: fresh reads `a,b,later` after Push or CommitBatch, and `later` after Delete. |
| Directory `PushAll` (235) | same | Runs as detached. Push then PushAll and Delete then PushAll disagree. |
| Directory `JudgeChecks` (269), `RefuseRemovalOfUndeclared` (259) | same | Run as detached. Agree. |
| In-body reads: `StoredAtOver` clock (4028), refusal text and `DurableAt` (4531, 4543), declares and cacheless refusal (4588, 4603, 4611), `Wrote` (4661) | `Cache`, under the gate `Gate` chose | On the attaching thread they read null while the thread holds the cache's gate, so a held call runs as a cacheless writer. |
| `AddBackingStore` (2259-2314) | writes `Cache` under both gates, after the pull | The window's author: the store is attached for the pull (registry lent, `Loaded` set, in `_stores`) while `Cache` is null. |
| Mesh `CultMeshSingleFileDocuments` writes and reads (99, 143, 174, 226) | none of the attached store's | `StoreAt` (239) builds a new store per call, so they never touch an attached store object. A file it shares with an attached store is the file lock's and the write set's business. |
| `IsDirty`, `HasStagedChecks`, `ForeignRecords`, `LastSchemaMigrationReports`, `HeaderFor`, `HeaderForWrite` | none | Ungated values or static decisions. They read no attachment fact. |
So during an attach, cacheless entry points decide attachment from the hold, and every held
entry point, plus every in-body read, decides it from `Cache`, which is still null. These are
two facts.
**Probe of the chosen shape** (commit 4695390c on b2330b6d; `proto-full.log`). `AddBackingStore`
publishes `Cache` before the pull and takes it back in the same hold if the pull fails. `Gate`
becomes straight-line code: guess, then the store's gate, where `Cache` is settled, then the
settled cache's gate. Results:
- All 21 probe cases are refused. The cacheless writes get the re-entry text and every other
case gets "A journal entered the cache it journals". After the flush, the attaching cache
and a fresh cache both read `a,later`.
- Caching: the only failures are the two tests that pin the late publish:
`AnAttachLendsTheRegistryAndPublishesTheCacheAfterThePull` (asserts no cache during the
pull) and `ACallWhoseGateChangesTwiceUnderItThrowsInsteadOfLooping` (asserts the old message
text; the tampering reader still throws). The churn, queued-reader and failed-attach tests
pass unchanged.
- Mesh and Networking are green.
- The cultcache-ts suite, which builds and drives the C# InteropPeer, passes on this shape:
175 tests, 171 pass, 0 fail, 4 skipped. The C#, Rust and Python interop tests ran, the C#
edit-in-place one included (`proto-ts2.log`, commit f5f02053). Recipe `clone/r3ts.sh`:
image node:24-bookworm, apt python3-msgpack, dotnet-install for 10.0, rustup minimal.
**Shapes weighed (rationale).**
- (a) Publish `Cache` before the pull, as a616431e did. Chosen. What s2 found was not that an
early publish is unbounded. It was that `Gate` re-read `Cache`*unlocked* after a stale
guess, so a churn of failing attaches could hand it a new stale guess every pass. The fix is
where `Gate` reads on its second step. Under the store's gate, `Cache` is settled: an attach
publishes and takes it back inside one hold of that gate, and no other thread can hold that
gate during the attach. A settled cache is permanent, because there is no detach. So `Gate`
needs at most three acquisitions, with no loop and no counter: the guess's gate, the store's
gate, and then the settled cache's gate, which cannot miss. This deletes a fact rather than
reconciling two. During the attach the store *is* attached, so every entry point, cacheless
or held, and every in-body read sees one answer. A journal's call is then refused by the rule
journals already have (AddJournal: "any hold on this cache while a journal runs throws"),
exactly as it is after the attach. Prior art: `ConcurrentHashMap.computeIfAbsent` installs a
ReservationNode in the bin *before* running the mapping function, so a recursive update on
the same thread finds the owner and throws, and on failure the reservation is removed (see
`docs/research/2026-10-23-store-lock-user-code-prior-art.md` item 4 on main: the re-check
after locking, and the ReservationNode on the empty-bin path). The validate-after-acquire in
`Gate` is the same item.
- (b) Keep the late publish and make the attach's hold the one fact for every entry point.
Rejected. `InStoreHold` knows that this thread holds the store's gate, not *whose* hold it is.
Refusing every same-store re-entry kills `UserCodeInACachelessCallMayCallItsOwnStore` (r2's
R5), so (b) needs an "attaching" marker beside `Cache`. That is the second fact that s3 and
s4 rule out, plus every in-body `Cache == null` read would need to consult it as well.
- (c) Run no user code on the attaching thread under the store's hold during a pull: defer the
pull's journals to after the hold (open calls). Rejected as insufficient on its own. Getters
and constructors still run during the pull's decode, so the window stays open to any user
code that is not a journal. It also moves journal timing, which is a documented contract
("right after an admission is applied").
- Scar check: `git log -S` over CultCache.cs finds `store.Cache = null` and `MaxStaleGuesses`
only in ba658965 and 1468a83a. The early publish was not removed for a correctness reason of
its own: 1468a83a removed it to bound `Gate` and to lend the codec, and the codec loan is gone
(`pulled-variant-codec`). No settled read under the store's gate was built before. The
precedent recipe (no root filter) finds no same decision.
- Not a product fork. A journal's store call during an attach now throws, as it already does
after the attach. Cacheless writes on the attaching thread stay refused, with the same text.
Getters and constructors during the pull see an attached store, as they do during any held
call. It changes the late-publish clause of Defaulted ruling `variants:ruling:pulled-variant-codec`,
so Self defaults a superseding ruling (recommended label `cache-published-for-the-pull`)
before dispatch.
- Residual, not in r3. `Push`, `Delete`, `PushAll` and `CommitBatch` are public, so user code
that is not a journal (another thread, or a getter during any held call) can stage records
behind an attached store's cache. Soul s4 shows this with `other-thread-push`: file `a,b`,
cache `a`. That is true before, during and after an attach, so it is not the window's
two-fact problem. Closing it changes public API (a store's writes taken only from its
cache), so it is a follow-up and an operator question for Self, not part of this cut.
## Body facts: claimant-one-owner r2 (Imagination, 2026-10-27, base fc74cf6f)
Why a revision: Soul beat the same rule three times (s1 claimant choice, s2 compatible ids, s3 shape). Each time a
cacheless write lent a registry let the registry's entry for the written schema id decide ownership beside the entry
the caller passed. Two descriptions of one type reached `Owns`; each fix (h2 schema-id compare, h3 subset of compatible
ids) compared one more facet of the two and left the rest open. r2 removes the second description from the lookup.
### Where the two descriptions met (source read, fc74cf6f)
- `OwnsOnDisk` cacheless arm (`CultCache.cs:4725-4740`): with a lent registry it called
`lent.Owns(id, catalog, declared, written)`, whose lookup is the registry's `_indexes` and whose `entryOf` is
`descriptor.ToCatalogEntry()` (`:824-825`). The claimant for the written schema id was the registry's descriptor,
so `ReadsWhole(persisted, claimantEntry)` (`:848`) judged shape against the registry's entry, and the claimant
precedence (`:756-806`) walked the registry's compatible ids. `written` (`:4734-4736`) then compared the two entries
by schema id plus a subset of compatible ids.
- `WriteDocument` (`:4495-4524`) takes its entry from the registry (`resolvedRegistry.GetRequired<TDocument>().ToCatalogEntry()`),
so its two descriptions are always identical. Only `WriteRecord(record, catalogEntry, holdsIds, registry)` (`:4471`) can
pass an entry that disagrees with the registry's entry for the same id.
- `CopiedHoldIds` (`:4600-4630`) also reads `AttachedRegistry`, to decide the v3 header from copied records. It is a
positive-only header decision, not ownership, and out of this cut.
### Consumer audit (2026-10-27)
- `WriteRecord` and `WriteDocument` are new in the Unreleased 3.0.0 entry; `main` has neither (`git show main:src/GameCult.Caching/CultCache.cs`).
- Per-repo `rg` over every git repo directly under `F:\Projects` for `.WriteRecord(`, `.WriteDocument(`, `WriteSingleFileDocument`
(Library, obj, bin, PackageCache excluded): hits only in CultLib and in vendored CultLib copies under Heimdall and StreamPixels
(`vendor\CultLib`, older pins without these methods). No consumer outside CultLib calls either method.
Mesh's raw-payload path calls `WriteRecord(..., registry: null)` (`CultMeshSingleFileDocuments.cs:129, 226`). The only
`WriteRecord` calls with a registry are tests: `CachelessSingleFileTests.cs:453, 660` (a registry `{IdDeck}` that does
not hold the written entry `One` at all, used for the copied-record header), `RestampOnWriteTests.cs:595, 599`, and
Soul's probes. Packages: no TS, Python or Rust counterpart takes a registry (cultnet-rs `write_record` is an example helper).
- Consequence: the registry's documented role in `WriteRecord` is to describe the other records in the store. A shape
that derives the written entry from the registry (or refuses an entry the registry does not hold) breaks
`CachelessSingleFileTests.cs:660` and removes that role.
### Candidate shapes weighed
| Shape | One description? | Public API | Kills s2/s3 | Cost |
|---|---|---|---|---|
| A. Lent registry's entry decides; refuse a passed entry that disagrees | yes, after a door check | signature kept; new refusal | yes | an equality check over a redundant parameter; breaks the registry-describes-others use (`:660`) when the registry lacks the written type unless it falls back to the passed entry, which is two descriptions again |
| B. Passed entry is the only description of its schema id; the registry supplies other types only | yes, by construction | signature kept; doc only | yes (probed) | one lookup class gains the registry; one registry overload deleted |
| C. Overloads: `WriteRecord(record, catalogEntry, holdsIds)` and `WriteRecord(record, registry, holdsIds)` | yes, by type | changes a 3.0.0 signature | yes | loses the registry-describes-others use |
| D. Delete `WriteRecord`'s registry parameter | yes | changes a 3.0.0 signature | yes | loses the copied-record header decision for raw writes |
Picked B. It is the only shape that keeps every documented use of the registry, needs no public signature change,
and deletes the comparison instead of extending it. It is ordinary scoping (an inner binding shadows the outer one
for its name), applied to schema ids.
### Probes (Starfire, local filtered runs, 2026-10-27)
Scratch clone of fc74cf6f at `<scratchpad>/imagination-claimant-r2/clone`, Soul's `SoulClaimantProbe.cs`,
`SoulClaimant2Probe.cs`, `SoulClaimant3Probe.cs` copied into `tests/GameCult.Caching.Tests`.
Base fc74cf6f (`probe-base.log`):
- P7 (registry {N lists own, A, B}; passed entry N with the list varied; record k under O, named like N): lent R / no
registry C for rows b, b2, e, g, h, k, l, l2; lent C for a, c, d, f, i, j. Confirms `subset-refuses-own-drift`.
- P9 (registry {NW: Name, Value, Colour}; passed NW minus Colour): lent C (restamped, Colour lost), no registry R.
Confirms `lent-shape-widens-passed-entry`.
- P6 (bare X1 entry, registry {X1} listing OX, record under OX): lent R, no registry R, cache C. h3 closed the id axis.
- P4: 12 states x 8 registries, violations=0. P1d: a cache {N,L} commits N over L's listed old id (`cache-replaces-other-types-record`).
Prototype B (scratch commit 349ea3f3 on the clone, not for landing; `probe-protoB2.log`): `CatalogEntryLookup` takes
the written entries and an optional registry and drops every registry descriptor whose schema id a written entry has;
the registry's `Owns(schemaId, catalog, declared, mayClaim)` overload is deleted; the cacheless arm always asks
`CultDocumentRegistry.Owns(id, catalog, declared, new CatalogEntryLookup(writes, AttachedRegistry), entry => entry,
claimant => writes has claimant's schema id)`.
- P7: lent == no registry on all 14 rows (all C).
- P9: lent R, no registry R, wide entry C.
- P6: unchanged (R, R, C). P4: identical table to base, violations=0. P1a-f, P2a-d: identical to base.
- `RestampOnWriteTests`, `CachelessSingleFileTests` and the probes: 139/139 passed locally.
| **Game definition** (documents, systems, authority declaration, tick rate) | Proposed: one CultCache document per server process (`gamecult.mesh.game_definition`), read-only to everyone but the server's startup. | Game id string from code (`"orbs"`), namespaced by Verse id. The id alone is not injective (two builds share it), so proposed identity is `(game id, definition hash)`, the hash derived from the lowered definition: document schema ids and versions, system ids, order and marks (`[Predicted]`, later `[Adjudicable]`), tick rate, authority declaration. Schema ids (`orbs.player`) are global strings; the game prefix is convention, not enforced (body fact F9). | Created at server startup from code; never revised at runtime; a new build supersedes it with a new hash. Clients carry their own compiled hash (C# in wasm, ruling `browser-sim-runtime`) and are refused at join on mismatch. A declaration a mode cannot honour is refused at startup (`modes-do-not-leak`). Never replayed. | The game's code, through the builder, at startup; the runtime validates. Forbidden: clients, any runtime mutation, Odin (Odin may read and index it, never write it). |
| **Game document** (Player, Pellet, Shot) | CultCache documents in the room store. | `(schema id, record key)`. Today the key is a free string the caller picks; `[CultIndex] Id` in the illustration is an index, not the key. Proposed: the runtime mints keys. Player key derives from the session principal (one Player per principal per room). Spawned documents (Shot, Pellet) key on `(spawner key, tick, spawn ordinal within that tick)`, so a client's predicted spawn and the server's authoritative spawn derive the same key and reconcile instead of duplicating. That is injective as long as spawn order within a system step is deterministic for the predicting client, which the predicted system's own code gives under one runtime (C# both sides). The developer's `Id` field becomes redundant; whether to keep it as a mirror of the key is an API call for the cut, not an operator fork. | Created by a system during a tick, or by the join system (Player). Revised at most once per tick, in that tick's commit. Deleted by a system (pellet eaten) or by the game's leave system after a session ends. Across reconnect within the grace window the Player record is unchanged; only the session's connection binding moves. Across a server restart: gone (ruled `room-durability`: ephemeral; a restart is a new room epoch at tick 0). Replayed only on the client, during reconciliation; the server never replays in stage 1. | The server tick runner on the room's primary is the one writer of committed state, and systems write only through the tick's working set. Forbidden writers: client document puts through `CultNetDatabaseServer.HandlePutAsync` (today accepted from any peer, F1); `PutPredictedAsync` and `CultMeshGameSession.PredictAsync` on any networked store; game code calling database APIs; the simulation-fact committer (F7). |
| **Player input** | Transient. A CultCache-encoded payload on a CultNet message; never a store record in stage 1. | `(session principal, input sequence)`, the sequence monotonic per session; each input also carries the tick it targets. Namespace: the room. Injective given the principal. The input type is a schema'd CultCache type, so the schema catalog publishes it like any document. | Sampled each client tick (devices read by the TS shell, handed to wasm); sent with the last few unacknowledged inputs repeated, so one loss is absorbed; buffered by the server per session against its target tick; consumed exactly once by the matching tick (proposed policy for late inputs: applied to the next tick and counted); acknowledged in the snapshot as "last processed sequence"; kept by the client until acknowledged, for replay; then discarded. Stage 4 changes this: inputs become signed, hash-chained and retained for a dispute window. | The client's principal authors the content; the tick runner decides which tick consumes it. Forbidden: server code altering input content; other clients; any input that arrives without a bound session. |
| **Tick** | Transient counter; proposed: the last committed tick is written into the room record in each tick's batch, so a snapshot or checkpoint knows it. | `(room id, room epoch, tick number)`, the number starting at 0 when the room epoch starts. Injective with the epoch. | Advanced only by the server tick runner at the one declared rate (ruled `tick-model`: one-fixed-rate); never revised. The client estimates the server tick from snapshots and runs its predicted tick ahead by its round trip plus a buffer; on reconnect it re-syncs from the next full snapshot. | Server tick runner. Forbidden: client clocks, transport arrival time, wall-clock reads inside systems (systems receive the tick, not the time). Today no tick scheduler exists anywhere in the C# body (F6). |
| **Predicted state and reconciliation** | Client-side only. In the browser they live in the C# wasm CultCache store; the TS renderer reads a per-frame view (ruled `browser-client-store`: wasm-owns-store). | Same `(schema id, record key)` as the authoritative document, plus the client tick it was predicted for. | Each client tick: predicted systems run over owned documents and the client's own spawns. On a snapshot for tick T acknowledging input sequence S: predicted documents are replaced by the authoritative ones at T, then inputs after S are replayed through the predicted systems up to the current client tick; predicted spawns are matched by derived key, and an unmatched one is removed once the acknowledgement passes its tick. Today `Reconciled` means "the authoritative put replaced the predicted copy" with no replay (F3); the replay loop does not exist. | Predicted state owns nothing outside the client. It is derived from the last authoritative snapshot plus unacknowledged inputs plus the predicted systems. The snapshot owns truth. Forbidden: sending predicted documents to the server; `CultNetClientAuthorityScope` in server mode (F4); the renderer writing. |
| **Session and the player's identity** | Proposed: session membership as a small CultCache document in the room store (`gamecult.mesh.game_session`, low churn, readable by systems and present in snapshots). The connection handle is transient process state. | Principal: the fingerprint of a browser-held P-256 key, proved by signing a join challenge (ruled `player-identity`: browser-keypair). Connection: a server-minted id per transport connection (ephemeral). Logical session: `(principal, room)`, bound across reconnects. Today the client's identity is a self-asserted `runtimeId` string (browser client options; `SourceRuntimeId` on operation envelopes); CultMesh's session identity proves the server to the client and not the reverse (F5). The body still has no authenticated client principal (F5, F13); the session cut builds it. | Connect, then join (definition hash check, principal proof), then the session record is created and the join system spawns the Player. Disconnect suspends the session for a grace window; resume binds a new connection to the same session; expiry ends the session, systems see the leave, and the game's leave system removes the Player. On server restart sessions end (ruled `room-durability`). | The server's join and leave handler on the room primary decides membership; the principal proves itself by signing the join challenge. Forbidden: a client-asserted `runtimeId` as identity; transport peer address; peer catalogs. |
| **Authority declaration and the lease it lowers to** | Declaration: inside the game definition. Lease: proposed as a CultCache document (`gamecult.mesh.authority_lease`); today a plain class in an in-memory dictionary (F2). | Declaration: per document type, `(writer, checker, arbiter)`. Lease: `LeaseId`, carrying Verse, peer, roles, shard ids, issuer, validity window, epoch and signature (`CultMeshAuthorityLease`). Namespace: the Verse. | Declaration fixed per build. At startup `Authority.Server` lowers to: the server holds a `ShardPrimary` lease on the room shard, and clients hold no write scope at all. Renewed before expiry; revocation or expiry stops the tick runner from committing; a new primary takes a new epoch. Proposed issuer in stage 1: the server's own deployment key, which clients pin through the browser trust policy, so an indie deployment needs no Odin. Odin-issued leases stay possible. | The game definition decides the declaration; the issuer issues leases; `CultMeshAuthorityResolver` decides whether a peer holds a role on a shard. Today the database's write gate (`CultNetShardDescriptor.IsPrimary`, `CultNetClientAuthorityScope`, both set from options) never consults the resolver: two authority structures (F2, F4). Proposed: the write gate is derived from lease resolution. First step (cut `remote-write-authority`): a peer's put or delete reaches the database only when `CultNetDatabaseServerOptions.RemoteWriteAuthority` grants it; none configured refuses every remote write (F1, GP1). The lease-backed grant needs an authenticated principal bound to the connection, so it follows the session cut. `CultNetClientAuthorityScope` is no longer an owner; it is derived from the declaration, and in stage 1 it is empty. |
| **Shard / room** (stage 1: one) | `CultNetShardDescriptor` (configuration); the room's state is its CultCache store. | Room id `(Verse, game id, room name)`, for example `orbs/main`, carried as the shard id plus the shard epoch. | Created at server start (epoch 1; on restart a new epoch, ruled `room-durability`); never moved in stage 1; ended at shutdown. Join means "join the one room"; matchmaking is out of scope. Interest management is stage 5 per the target, so in stage 1 every client receives the whole room. | The runtime holding the `ShardPrimary` lease. Forbidden: a second primary; clients. |
| **Snapshot sent to clients** | Transient. A CultNet message whose payload is CultCache-encoded documents; never stored. The server keeps per-client baselines in memory for deltas. | `(room, epoch, tick, recipient session, baseline tick)`. | Derived after a tick commit at the send rate; a delta against the recipient's last acknowledged snapshot, or a full snapshot on join and resume; carries the last processed input sequence for that session. Discarded once sent; a baseline is dropped when a newer one is acknowledged. Today the nearest things are per-record change pushes from `CultNetDatabaseSubscriptionServer` (no tick, no acknowledgement, no input sequence) and shard catch-up snapshots in `CultMeshSnapshots.cs` (F8). | The snapshot builder decides nothing; it is derived from the committed tick. Forbidden: the per-record subscription stream acting as a second game replication path. That stream stays for non-game observers. |
| **Simulation fact / witness observation** (found) | `CultMeshSimulationFact` is a CultCache document; observations are hub memory. | `(shard, epoch, frame, subject, kind)`. | Stage 1: not used. The target replaces it at stage 4. | Today `CultMeshGameSession` serves observations by default (`ServeSimulationObservations = true`) and commits quorum facts through the primary (F7). In server mode there is no checker, so the observation server and the committer must not be wired. Forbidden writer in stage 1. |
| **Shard mutation log entry** (found) | `CultNetDatabase` in-memory `SortedList` per shard, plus an optional durable store. | `(shard id, sequence)`. | Today every primary write appends one entry, and the in-memory list is never trimmed (F10). Per-record writes at tick rate would grow it without bound. In stage 1 the tick's write set either bypasses the per-record log or lands as one bounded entry per tick. Ruled: the tick's write set lands as one batch per tick (`tick-state-commit`), in memory (`room-durability`); the bounded per-tick entry is the tick-runner cut's to specify. | `CultNetDatabase` on the primary. Forbidden: per-system-write appends from the tick runner. |
### Cells filled by rulings (2026-10-27)
- Session principal: `player-identity` (browser-keypair). Not yet in the body (F5, F13).
- Browser documents and the renderer's read: `browser-client-store` (wasm-owns-store).
- Room state across restarts: `room-durability` (ephemeral).
- Tick rate model: `tick-model` (one-fixed-rate).
- How a tick reaches the store: `tick-state-commit` (tick-batch; the tick-runner cut times `CultCacheBatch` at tick rate first).
- Stage 1 transport: `browser-transport` (WebSocket behind a port carrying only inputs and snapshots; a WebTransport server in parallel, `webtransport-server-layer`: managed C# over the MsQuic bridge).
- Still unprobed: `CultCacheBatch` commit cost at tick rate for about 50 players plus pellets (wasm measuring probe and tick-runner cut).
## Body facts
All are source reads at CultLib `d1bf21b3` on 2026-10-27 by `imagination-gp-0b`. No
runtime probe was run in this pass. Each fact names the read that would falsify it.
- **F1. The primary applies any peer's document put.**`CultNetDatabaseServer.HandlePutAsync(message, peer)` (`src/GameCult.Networking/CultNetDatabaseServer.cs:307`) calls `ApplyPutAsync(message)` (`CultNetDatabase.cs:1022`), which checks only that this runtime is primary for the shard and the epoch matches (`EnsurePrimary`, `:2003`). `peer` is used only to send errors. Any connected client can overwrite any game document on the server.
- **F2. Leases are not typed state and do not gate writes.**`CultMeshAuthorityLease` is a plain class (no `[CultDocument]`); `CultMeshAuthorityLeaseCatalog` is a `Dictionary` (`CultMeshAuthority.cs:242`). The only `[CultDocument]` types in `GameCult.Mesh` are body publication, CDN chunk and manifest, content transfer state, discovery state, entity prefab package and simulation fact. `GameCult.Networking` (the database) references no lease type; the one Networking file naming `Lease` is `CultNetSchemaMessages.cs`.
- **F3. Reconciliation is replacement, not replay.**`CultNetDatabase.cs:1247-1252`: a predicted put records a prediction key; the next non-predicting write to that key turns the change into `Reconciled`. No code re-applies inputs after an authoritative state.
- **F4. Client write scope is a runtime-id string match from options.**`CultNetClientAuthorityScope.Matches` compares `OwnerRuntimeId` with the database's own `_runtimeId` (`CultNetDatabase.cs:111-153`, `EnsureClientAuthority``:1934`). It gates local prediction only and has no relation to leases.
- **F5. No authenticated client principal.**`CultMeshSessionIdentity.cs` verifies transports for `(verse, authorityRuntimeId, protocol, routeGeneration)`, which proves the server to the client. The browser client's `runtimeId` (`packages/cultmesh-browser/src/index.ts:182`) and the operation envelope's `SourceRuntimeId` (`CultNetOperationServer.cs:39`) are self-asserted.
- **F6. No tick scheduler.** A grep of `GameCult.Networking` and `GameCult.Mesh` for tick or timestep names finds none. `Frame` exists only as a witness-asserted number in `CultNetSimulationConsensus` and `CultMeshSimulationFact`.
- **F7. The game session serves the witness path by default.**`CultMeshGameSession` constructs `CultNetSimulationObservationServer` when `ServeSimulationObservations` (default `true`) and always constructs a `CultMeshSimulationFactCommitter` (`CultMeshGameSession.cs:70-104`). `PredictAsync` is `PutPredictedAsync` (`:161`).
- **F8. The browser can watch and invoke, not put.** The `cultmesh-browser` README says it "does not expose direct document writes". The client offers `leaseRawDocument` (`index.ts:349`) and `invoke` of typed operations (`:370`), over one WebSocket. On C#:`CultNetDatabaseSubscriptionServer` and `CultNetOperationServer` take `ICultNetSchemaServer`, so they run over `CultNetWebSocketSchemaServer`; `CultNetDatabaseServer` (puts) takes the concrete UDP `Server` (`CultNetDatabaseServer.cs:36`). Operations are request and response with receipts and idempotency keys, which is a poor fit for an input stream at tick rate.
- **F9. Schema-id namespacing is convention.**`[CultDocument("orbs.player", ...)]` ids are global strings; nothing ties a schema id to a game id.
- **F10. The in-memory mutation log is never trimmed.**`RecordMutationLogEntry` (`CultNetDatabase.cs:1610`) inserts into a per-shard `SortedList`; no `Remove` or trim on that list exists in the file. Compaction applies to the durable store and to burned sequences only.
- **F11. CultCache has a write-set batch; CultNetDatabase does not expose one.**`CultCacheBatch` with `Commit` and `TryCommit` (`GameCult.Caching/CultCache.cs:1796`, `:2212`) is the 2.0.0 write set (`variants:ruling:write-set-content-version-declared-ownership`). `CultNetDatabase` has no batch method: every write is a per-record door.
- **F12. A reflection site on the mesh put path.**`CultMesh.cs:3502``PutDatabaseDocumentAsync` picks `PutAsync` or `PutPredictedAsync` by `GetMethods` and `MakeGenericMethod`, which matters to the AOT prerequisite (`cultlib-gaps:follow_up:cs-aot-safe-registration`; probe in `research/cultmesh-garden-path/cultcache-aot-probe-2026-10-27.md`).
- **GP1 (runtime probe, 2026-10-28, `imagination-gp-stage1`).** Scratch NUnit test at `51ca81a1`, run on Yggdrasil (`ygg-verify.sh`, `mcr.microsoft.com/dotnet/sdk:10.0`, filter `ImagGpProbeTests`, never pushed to origin). The server writes `gp:player` = "server-owned" through `CultNetDatabase.PutAsync` on a primary shard. `CultNetDatabaseServer._putHandler`, invoked with an uninitialized `CultNetServerPeer` (no field set: no endpoint, no transport, no identity) and a put of "client-overwrote", leaves the record "client-overwrote", and `Server.CanProcessBeforeVerification(put)` returns `true`. Output: `GP1 after=client-overwrote canProcessBeforeVerification=True`, Passed 1. This confirms F1 at runtime: nothing about the peer is consulted, and puts and deletes pass before the legacy login verification (`Server.cs:562-575`).
- **F13. A server peer carries no identity.**`CultNetServerPeer` (`Server.cs:829`) holds the LiteNetLib `NetPeer`, its endpoint and transport stats. The only principal on `Server` is the legacy username and password login (`PlayerData.PlayerId`, `IsVerified``:697`), which the database door never reads. `ServerSecurityOptions` is one shared `ConnectionKey`: transport admission, not authority.
- **F14. A new refusal code is a coordinated change.** The `cultnet.error.v0``code` is a closed enum (`contracts/cultnet/cultnet.error.schema.json:28-31`, with a byte-identical copy under `packages/cultnet-rs/contracts/`). Rust `CultNetErrorCode` (`packages/cultnet-rs/src/contracts.rs:242-273`) refuses unknown codes, `packages/cultnet-rs/tests/error_message.rs` decodes captured C# bytes per code, and TypeScript imports the root schema (`packages/cultnet-ts/src/contracts.ts:10`). `unowned_schema` was added this way and is the pattern.
- **F15. Forwarding re-sends as a new client.**`TryForwardPutAsync` (`CultNetDatabaseServer.cs:654`) forwards a `not_primary` put through `ICultNetShardWriteForwarder`, and `CultNetSchemaWriteForwarder` opens its own client connection (`CultNetShardWriteForwarding.cs:122-128`). The primary therefore sees the forwarding runtime, never the original peer.
- **F16. Consumers of the remote write door.** No C# consumer outside CultLib constructs `CultNetDatabaseServer` (grep of the Aetheria, Epiphany, Ghostlight, huginn and gamecult-ops `*.cs` files finds only vendored CultLib copies). In CultLib, tests reach the put and delete handlers directly in `CultNetJournalTests.cs:738`, `:935-936` and `SoulP4cProbeTests.cs:57`. `CultNetDatabaseServer.ApplyPutAsync` and `ApplyDeleteAsync` (`:73`, `:81`) have no caller in `src` or `tests` besides the handlers. TypeScript puts (the `cultnet-ts` Idunn presence and signed health publishers) target Rust servers.
- **F17. The Rust document server admits a put by size only.** In `packages/cultmesh-rs/src/rudp_document_server.rs:682-720`, a `DocumentPutRaw` is refused only when its served response would not fit; no authority is asked. This is F1's gap in the Rust runtime (a follow-up, not stage 1).
## Cut sequence (stage 1)
One line per cut not yet mapped. The model page has no empty cell as of 2026-10-28. Remote write authority (`cultmesh-garden-path:cut_spec:cut-remote-write-authority.r1`, mapped 2026-10-28, so no longer a line here) goes first: it waits on nothing, every later server cut stands on a closed write door, and GP1 shows the door open today. The wasm measuring probe is the sibling that may run beside it, since it touches no file this cut does; ruling `browser-sim-runtime` names it the first cut of the browser runtime track, not of stage 1 as a whole.
- **Wasm runtime measuring probe.** Download size, startup and 60 Hz tick cost of a stub orbs tick under Mono browser-wasm and NetWasm, including `CultCacheBatch` commit cost at tick rate. This is the first cut named by ruling `browser-sim-runtime`, which it can overturn. Its NetWasm and AOT variant waits on `cultlib-gaps:follow_up:cs-aot-safe-registration`.
- **Witness services off in server mode.**`CultMeshGameSession` stops constructing the observation server and fact committer when the declaration has no checker (F7). Waits on nothing; can run beside remote write authority.
- **Leases as typed state and the lease-derived grant.**`CultMeshAuthorityLease` becomes a CultCache document, the lease catalog a store view, and the Mesh host's `RemoteWriteAuthority` resolves the connection's principal through `CultMeshAuthorityResolver`. Waits on remote write authority and on session and principal.
- **Demote `CultNetClientAuthorityScope`.** It becomes derived from the authority declaration (empty in server mode). Waits on game definition.
- **Changelog entry for the remote write refusal.**`unity/org.gamecult.cultlib/CHANGELOG.md` is also written by `variants:cut_spec:cut-write-ownership-cs.r1`, so Self adds this cut's bullet at merge, after that cut lands, and the two specs share no file.
- **Game definition and server tick runner.** Builder, document and system registration, definition document and hash, fixed tick, tick-boundary commit. Rests on rulings `tick-model`, `tick-state-commit` and `room-durability`; waits on nothing mapped.
- **Session and principal.** Join, resume and leave; principal proof; session document. Rests on ruling `player-identity`; waits on game definition (the join checks the definition hash).
- **Input and snapshot path.** Input messages with redundancy, a server input buffer, acknowledgements, delta snapshots. Rests on ruling `browser-transport`; waits on session and principal.
- **Browser client runtime.** Wasm simulation host, TypeScript shell for transport and rendering, prediction with replay reconciliation. Waits on the measuring probe and `browser-client-store`.
- **Orbs reference game and garden-path proof.** The game written with no networking code (`garden-path-needs-no-networking`), and a browser client that writes (`browser-is-first-class`). Waits on all of the above.
@ -34,7 +34,7 @@ ran under the session scratchpad (`imagination/`), not in the repo.
| F13 | Cross-process precedent for C# against a Node provider exists. `packages/cultnet-ts/test/interop/cultnet-interop.test.ts:536-560` spawns the TS QUIC provider, then runs `dotnet test --filter NativeConnectorReceivesFromExternalProvider` with the endpoint in `CULTMESH_NATIVE_EXTERNAL_ENDPOINT`. The C# test skips without the variable (`tests/GameCult.Mesh.Quic.Native.Tests`, line 21). | source |
| F14 | `cultnet-ts` is CommonJS. Its `./wire`, `./contracts` and `./authority` subpaths import only `@msgpack/msgpack`, `ajv` and `cultcache-ts/document`, so they are browser-safe. Its root `index` pulls `node:` modules (`peer`, `rudp`, `transport`, `security` and others). `cultmesh-browser` (ESM) already imports `cultnet-ts/wire`, `/contracts` and `/authority`. `cultmesh-ts` loads `koffi` lazily, inside a function (`realtime-quic-native.ts:139`), so a new subpath of it pays no native load. The root has `esbuild` ^0.28.1 as a devDependency. | source, grep |
| F15 | TS packages build with `tsc` and test with `node --test dist-test/test/*.test.js` per workspace. `scripts/run-typescript-workspaces.mjs` runs them in order `cultcache-ts, cultnet-ts, cultmesh-ts, cultmesh-browser`, after `check-typescript-tests-wired.mjs`. No StrykerJS config exists in CultLib (`docs/mutation-testing.md:156-159`). | source |
| F16 | **No CultLib TS package except `cultcache-ts` has a publish job.**`publish-packages.yml` triggers on `cultcache-ts-v*`, `cultcache-py-v*`, `cultnet-py-v*` and `cultmesh-py-v*`only. Tags `cultnet-ts-v0.1.1` and `cultmesh-ts-v0.1.1` exist, but no workflow consumes them. `npm view` returns 404 anonymously for `cultnet-ts`, `cultmesh-ts`, `cultmesh-browser` and even `@gamecult/cultcache-ts` (`@gamecult/cultcache-ts` may be private; unverified). The doctrine's rule of at most three tags per push stands (SKILL.md:674). | source; `npm view` |
| F16 | **No CultLib TS package except `cultcache-ts` has a publish job.**`publish-packages.yml` triggered on `cultcache-ts-v*`, `cultcache-py-v*`, `cultnet-py-v*` and `cultmesh-py-v*`when this was written; it now triggers on the three `-py-v*` patterns only, because npm publishing is deliberately off until ruled on (a tag run uses the workflow at the tagged commit, so `cultcache-ts-v0.15.0` on `7d292085`, before that change, still ran the old one). Tags `cultnet-ts-v0.1.1` and `cultmesh-ts-v0.1.1` exist, but no workflow consumes them. `npm view` returned 404 anonymously on 2026-11-08 for `cultnet-ts`, `cultmesh-ts`, `cultmesh-browser` and even `@gamecult/cultcache-ts` (`@gamecult/cultcache-ts` may be private; unverified). The doctrine's rule of at most three tags per push stands (SKILL.md:674). | source; `npm view` |
| F17 | StreamPixels consumes CultLib as the git submodule `vendor/CultLib` (gitlink `30ee8b97`) with `file:` deps and `pnpm.overrides`. Its CI and Idunn recipes build the vendored workspaces (`npm ci --ignore-scripts`, then build `cultcache-ts`, `cultnet-ts` and `cultmesh-ts`). `apps/web` depends only on `cultnet-ts` today, not on `cultmesh-browser`. | StreamPixels `.gitmodules`, `apps/*/package.json`, CI workflow lines 62-95 |
| F18 | The Rust reader sets two guards that the TS reader should match. A caller cap `max_bytes` is checked before any request. And a zero-size chunk is refused unless it is the only one, because a manifest of zero-size chunks would be a request flood. | source `cultnet-rs/src/content.rs:364-385` |
| F19 | `BuildChunkUri` on more endpoint shapes, as `AbsoluteUri` (what `HttpClient` sends). `https://h` and `https://h/` both give `https://h/?chunkHash=…&recordKey=k`. `https://h:443/x` drops the default port. `https://H.Example/X/` lowercases the host and keeps the path's case. `https://h/x#frag` keeps the fragment after the query. `https://h/x?a=%20b+c` keeps the existing query verbatim and joins with `&`. **`https://h/x?` (an empty query) gives `https://h/x?&chunkHash=…`**, because `UriBuilder.Query` is `?`, which is not blank. `https://[::1]:8443/c` keeps the bracketed host. `Uri.EscapeDataString("!'()*~-._ +é𐐀")` is `%21%27%28%29%2A~-._%20%2B%C3%A9%F0%90%90%80`: unlike JS `encodeURIComponent`, it escapes `!'()*`. | probe: `dotnet fsi` on .NET 10, the three lines of `BuildChunkUri` (`CultMeshHttpsContentTransport.cs:114-122`) copied verbatim |
@ -276,11 +276,10 @@ remains:
history (Keep a Changelog headings).
- A Soul pass over the release diff, then four tags: `cultcache-ts-v…`,
`cultnet-ts-v0.3.0`, `cultmesh-ts-v0.2.0` and `cultmesh-browser-v0.4.0`.
These go in **two pushes** (at most three tags per push, SKILL.md):
`cultcache-ts` first, because its tag triggers `publish-packages.yml`
(F16), and the three plane tags second. Whether the `cultcache-ts` tag's
npm publish job still runs, and to where, is checked before that push;
follow-up `npm-publish-ts-packages` owns the rest.
These go in **two pushes** (at most three tags per push, SKILL.md). No
`-ts-v*` tag triggers `publish-packages.yml` on main (F16), so these tags
@ -441,8 +441,8 @@ consumer that waits wants delivered, which is now just Acknowledged.
| `cultmesh-rs/src/lib.rs:781-838``request_raw_snapshot_from_rudp_catalog` (Idunn `src/drivers.rs:5527` uses it) | a hand-rolled loop until `Acknowledged`, then the response | Cut 4: `wait_acknowledged` |
| `cultmesh-rs/src/lib.rs:870-900``publish_cultnet_message_to_rudp_catalog` (Odin `odin-daemon`) | a hand-rolled loop until `Acknowledged`, then disconnect. The doc at `:869` claims "application admission" | Cut 4: `wait_acknowledged`; the doc says delivered to the document server's sink (`rudp_document_server.rs:343-372`) |
| `cultmesh-rs/src/lib.rs:918-938``publish_cultnet_messages_to_rudp_catalog` (Ghostlight `mesh.rs:323`) | `flush_reliable`, then drop. **Several ordered frames, then close** | None; flush now means it |
| `cultmesh-ts/src/index.ts:6070-6098` `publishRudpDocumentOnce` (Heimdall, StreamPixels, Stonks, weksa, Bifrost, VoidBot, Hermodr) | `peer.flush` (`cultnet-ts/src/peer.ts:109-110`), then close | None. Cut 5 moves its connect-and-wait helper (`index.ts:6603-6650`) into cultnet-ts |
| `cultnet-ts/src/idunn-odin-presence-publisher.ts:89-215` `publishDocument`, behind `createIdunnRuntimePresencePublisher`. **StreamPixels' service runs this**: `apps/service/scripts/idunn-entry.mjs` runs `deployment/idunn/runtime-presence.mjs`, which calls `createIdunnRuntimePresencePublisher` (`:46`) from its vendored CultLib `542ddd3` (StreamPixels `a49f7d8`; pass 2's `30ee8b9` is stale), where it lives under its older name `cultnet-ts/src/idunn-runtime-authority.ts:206` | a hand-rolled session loop (`:186-208`) until `outstandingReliablePacketCount === 0`. It does resend (`:205`). It drops `readyToSend` (`:198`), so promoted queued packets wait for the resend timer. It pins `initialSequence: 1` (`:103`) and never sends a Disconnect, so Odin's hub keeps each publish's peer until it times out (each publish binds a new port, so each is a new peer) | Cut 5: the socket transport, `connectAndWait`, `send("schema")`, `flush`, `close` (which sends the Disconnect, `rudp.ts:956-969`) |
| `cultmesh-ts/src/index.ts:6211-6303` `publishRudpDocumentOnce` (Heimdall, StreamPixels, Stonks, weksa, Bifrost, VoidBot, Hermodr) | `peer.flush` (`cultnet-ts/src/peer.ts:109-110`), then close | None. Cut 5 moves its connect-and-wait helper (`index.ts:6603-6650`) into cultnet-ts |
| `cultnet-ts/src/idunn-odin-presence-publisher.ts:97-136` `sendPresenceDocument` (formerly `publishDocument`), behind `createIdunnRuntimePresencePublisher`. **StreamPixels' service runs this**: `apps/service/scripts/idunn-entry.mjs` runs `deployment/idunn/runtime-presence.mjs`, which calls `createIdunnRuntimePresencePublisher` (`:46`) from its vendored CultLib `542ddd3` (StreamPixels `a49f7d8`; pass 2's `30ee8b9` is stale), where it lives under its older name `cultnet-ts/src/idunn-runtime-authority.ts:206` | a hand-rolled session loop (`:186-208`) until `outstandingReliablePacketCount === 0`. It does resend (`:205`). It drops `readyToSend` (`:198`), so promoted queued packets wait for the resend timer. It pins `initialSequence: 1` (`:103`) and never sends a Disconnect, so Odin's hub keeps each publish's peer until it times out (each publish binds a new port, so each is a new peer) | Cut 5: the socket transport, `connectAndWait`, `send("schema")`, `flush`, `close` (which sends the Disconnect, `rudp.ts:956-969`) |
| `cultnet-ts/src/signed-daemon-health.ts:104-162``publishSignedDaemonHealth` (Hermodr, Odin `idunn-rudp.cjs`) | a hand-rolled loop: sends the packets `sendMany` admitted, waits up to 500 ms for the **first** Ack packet of any content (`:146-153`), swallows the timeout, closes. Resends run inside that wait (`:303`) but promoted queued packets are never sent | Cut 5, as above. Today a write not acked within 500 ms, or any payload over 32 packets, is reported published |
| `cultnet-ts/src/operation-service.ts` | the operation response | none (the response is the proof) |
- Shapes the walk does not own already behave as whole values. A keyed document holding
`ImmutableArray<PElem>`, `(List<PElem>, int)`, `KeyValuePair<string, List<PElem>>`, a non-enumerable
`[DataContract]` holder, or a string-keyed holder (`[Key("items")]`) gives `Inspect` problems 0, an empty
IdPath, `CanHoldIds` false, `Plan` 0 (random and deterministic), and `Holds` false with ids set.
- The marker and the copy check do not count them. Committed with ids `aaaaaaaaaaaa`/`bbbbbbbbbbbb`, the store
header stays `cultcache.store.v1`. A fresh cache that writes another record over that file (the copy-forward
check) keeps `v1`. Reload returns the ids exactly as written, and unset ids stay null after write and reload.
- Variants replace them whole, and that is the C2a resolver's behaviour for every list. A variant `Set` of the
`ImmutableArray` slot resolved to `[9:eeeeeeeeeeee]` with the base untouched, header `v2`, `HoldsIds` false.
A path step `[1:aaaaaaaaaaaa]` is refused: "this runtime resolves top-level Set overrides only". No C2a gap.
The gap belongs to C2b, in the cut sequence below.
- AllowPrivate is read from the most derived `[MessagePackObject]`. `ApDerived` (AllowPrivate over a base
without it) is written `[[],[],4,[]]`: base-chain privates and a private-getter property, with
`ObjectSerializationInfo` at `allowPrivate: true` = `[baseStash, own, N, Pg]`. `NoApDerived` (plain over an
AllowPrivate base) is written `[nil, 5]`. The walk read only `[N]` for both.
- A `[DataContract]``List<T>` subclass is written by `NonGenericListFormatter`, and its record does not read
back ("System.Object[] is not of type PElem"), whatever the walk does. That is a MessagePack limit, not an
id fact.
## 2. Model page (step 0b)
| Kind | Named by | Over time | Decided by |
@ -99,16 +214,16 @@ Probes run 2026-10-12 from `F:\Projects\CultLib-c2afix` with git only (no builds
| Base reference | the base's key, in the record's variant slot | changed only by an explicit rebase; a base cannot be deleted while it has variants, unless the same batch rebases or removes them | author; the cache refuses a cycle, a missing base, a different concrete type, a global type |
| Override entry | `(op, path, id, value)`; op is Set, Insert or Remove | added by editing; removed by clearing; survives base edits while its path resolves | author; the cache refuses entries a base edit strands (Q2); an entry whose member left the type soft-drifts with a warning (Q4 c) |
| Member path | a sequence of `(slot, elementId)` steps; `slot` is the MessagePack `[Key]`; `elementId` is `""` unless the step enters an object-list element | a type change can invalidate it | the type's owner (slot authority); ids by the element's creator |
| Element id | a `[CultElementId]` string member on an object-list element type; `""` means unset | minted by the cache on write when unset; copied unchanged into variants; a moved element is a new element | cache mints it; it is never retargeted |
| Element id | a `[CultElementId]` string member on the element type of an object list in a shape the walk owns (arrays, `IEnumerable<T>`, `IDictionary`/`IReadOnlyDictionary` values, `[MessagePackObject]`/`[Union]` types with integer keys), one MessagePack writes and reads back (integer `[Key]`, no `[IgnoreMember]`; public, or any visibility under the most derived `AllowPrivate`); `""` means unset. **Exception (q3 amended 2026-10-16):** a shape the walk does not own is one whole value: `ImmutableArray`, `ArraySegment`, `Memory`, `KeyValuePair`, tuples, `Lazy`, unkeyed or `[DataContract]` classes, string-keyed types, a type or member with `[MessagePackFormatter]`, and a non-public member without `AllowPrivate`. Its elements carry no ids: a `[CultElementId]` inside is a plain string | minted by the cache on write when unset; copied unchanged into variants; a moved element is a new element. Inside a whole value: never minted, never checked, never counted for `v3` | cache mints it; it is never retargeted. `CultElementIds.Classify` alone decides owned versus whole |
| Resolved view | `CultStoredDocument.Document` for a variant key | recomputed inside the hold that admits the variant, or any base on its chain | the cache only; never persisted, never written back (Q1) |
| Index and name entries | as today, from the resolved view | a dependent's entries are dropped and re-added in the same `Apply` as its base change | the cache; a variant must override the `[CultName]` member (rule R6) |
| Watches | as today; each re-resolved dependent emits `Updated` with its own `Sequence` | published in the same hold as the base change | the cache |
| Compare-exchange | `(schemaId, storedAt)` of the stored delta (ruling 2) | a base change does not stamp its variants | the cache; the resolved view is not an identity |
| Persisted format version | store header `formatVersion` | `cultcache.store.v2` when the store holds at least one variant; `v1` otherwise (Q5) | the writing store, from content |
| Persisted format version | store header `formatVersion` | `cultcache.store.v2` when the store holds at least one variant; `v1` otherwise (Q5); a marked `v3` (`v5` directory) stays while a write copies any record forward, and only a write that encodes every record drops it by content | the writing store, from content; the durable header alone says the bytes hold element ids |
| Single-file store file | the path; one MessagePack array `[header, catalog, records]` | replaced whole and atomically; refused when unreadable; see "Store authorities" | each runtime's one reader decides what is a store; the writer derives the header from it |
| Store lock and staging files | `<file>.lock`; a unique temp beside the store | the lock is taken by C# and Rust writers (CultMesh included, through `ReplaceDurable`); staging is removed on a failed write | the store, never CultMesh or a consumer |
| Variant slot on the wire (CultNet) | **empty cell**: no CultNet message carries a delta today | C5 | ruling 4 decides deltas; the message shape is C5's design |
| Dictionary entries, scalar lists, arrays | **empty cell**: no stable identity exists | whole-member Set only (Q7) | Q7 |
| Dictionary entries, scalar lists, arrays, whole-list holders | **empty cell**: no stable identity exists | whole-member Set only (Q7; whole-list holders by q3 amended) | Q7, q3 amended |
| Template (non-item) bases | **empty cell**: parked by ruling 6 | none | not in these cuts |
## 3. Where the Body contradicts the target
@ -220,20 +335,21 @@ Dictionary entries, scalar lists and arrays are replaced whole.
- **Recommend yes.** Aetheria's cases (`Hardpoints` whole; `Behaviors[id]...Max`) fit. Keyed
dictionary addressing can be added later without a format change. Affects C2.
## Cut sequence (2026-10-12)
## Cut sequence (2026-10-15)
Mapped now (specs in the mind): `c2a-declared-apply` r2, the C# commit-time fix batch on
`hands/variants-c2a` (the stash, finished, with tests and contract text), and its sibling
`main-into-c2a` r2, main merged into C2a on `hands/variants-c2a-main` in its own worktree. Later
cuts, one line each, mapped only when Self is about to dispatch them:
Landed since 2026-10-12: `c2a-declared-apply` (r3, Soul s5) and `main-into-c2a` (r2, Soul s2); the
reconcile step is done, `hands/variants-c2a` is at the reconciled head `6afa8ed9`. Element ids part 1,
`c2a-element-id-holes` (r6), is held by Soul s5/s6 on `hands/variants-c2a-element-id-holes`. Mapped
now: part 2, `c2a-id-closed-world` r2 (2026-10-16, under the amended q3), not blocked. Its open fork
`variants:question:c2a-whole-list-opt-in` holds under every option. Later cuts, one line each, mapped
only when Self is about to dispatch them:
1. **reconcile**: merge `hands/variants-c2a` into `hands/variants-c2a-main` after both Soul passes
hold; Self moves `hands/variants-c2a` there. Waits on both mapped cuts.
1. **reconcile**: done at `6afa8ed9`.
2. **parity-ts**: TypeScript write set, one record per key, undeclared records carried and refused,
unheld removals judged at commit. Waits on reconcile.
unheld removals judged at commit. Waits on element ids parts 1 and 2 merged into C2a.
3. **parity-rs**: Rust `push_all` and `put_prepared_batch` apply a write set, ownership by
declaration, the same commit-time checks. Waits on reconcile; can run beside parity-ts.
4. **parity-py**: Python write-set apply. Waits on reconcile and on `cultlib-gaps` cut
declaration, the same commit-time checks. Waits as parity-ts; can run beside it.
4. **parity-py**: Python write-set apply. Waits as parity-ts, and on `cultlib-gaps` cut
`py-msgpack-default` landing on main (then a main merge into C2a first).
5. **write-set-vectors**: one cross-runtime manifest, read and acted on by all four runtimes;
@ -242,6 +358,11 @@ cuts, one line each, mapped only when Self is about to dispatch them:
with an empty diff against the verified head; no tag. Waits on write-set-vectors, Soul holding,
and `asura:cut_spec:cut-semver-measured` landed. Asura Cut 3 bases on its merge commit.
Element-id part 2: `c2a-id-closed-world` r2 decides deferred findings `s5.member-set-misses-allowprivate`, `s5.enumerating-rule-unpinned`, `s6.null-standard-untyped`, `s6.two-recognisers-one-rule`, `s4.unrecognised-holders-fail-open` and follow-up `c2a-datacontract-holder-unwalked` (see its `first`).
- **marker-for-whole-lists** (only if `c2a-whole-list-opt-in` is ruled `marker-when-ids-declared` or `marker-for-any-keyed-object`): a CarriesElements predicate, a refusal unless the member is marked, and a public attribute. It is additive after part 2.
- **C2b constraint** (owed to `variants:follow_up:remaining-variant-cuts`): nested path resolution must refuse a step `[slot:elementId]` that enters a whole value, naming the holder. A whole value is any shape `CultElementIds.Classify` calls Leaf, so C2b asks Classify and keeps no recogniser of its own. Today's C2a resolver refuses every non-top-level step, so no store can hold such an override yet.
Rationale. The 2026-10-02 order (parity, vectors, then merge main) was right when main was close.
Main has since moved 453 commits and rebuilt exactly what the parity cuts change: cultcache-ts
payload ownership, cultmesh-ts, cultcache-rs. Parity mapped before the merge would be mapped
@ -1484,7 +1605,7 @@ The one door per runtime, and who walks through it:
- **Inputs.** The bytes at the path. **Outputs.** The decoded store with header and catalog; or a typed refusal (`CultStoreUnreadableException` in C#; `CultCacheStoreUnreadable{path, kind}` in Rust, kind `Undecodable` or `UnsupportedFormat`; `StoreUnreadableError` in TS and Python). A refusal names a header only as `cultcache.store.v<digits>`.
- **Derived state.** The write header is never decided by the writer: it is the reader's verdict (header read, whole-store flag, whether ids or variants are held). A rewrite of a file that cannot be read cannot happen, because the reader throws first.
- **Forbidden writers.** A second header classifier. Deleted: Rust `leading_header`, `read_array_header`, `read_string`; cultnet-ts's own decode of the store header; the TS async `readStore`; Rust `StoreUnreadableError` (negative grep `StoreUnreadableError` in `packages/cultcache-rs`: no match); Python `_decode_v1_snapshot`.
- **Shared paths.** cultnet-ts reaches the header decision only through `readSingleFileStore` (AUTH:4, :321) and checks only record count, schema id and type. cultnet-ts therefore declares cultcache-ts `^0.15.0`, the first release that exports it.
- **Shared paths.** cultnet-ts reaches the header decision only through `readSingleFileStore` (AUTH:4, :321) and checks only record count, schema id and type. cultnet-ts therefore needs a cultcache-ts release that exports it, which is the release that ships with this change; the manifests move to that version then (c0-ships-with-c2a), not before.
- **Pinned by.**`tests/vectors/document-variants-c2a/readability/manifest.txt`, walked by `StoreReadabilityTests.OpenReadsExactlyTheFilesTheVectorsSayItReads` (C#), `CultMeshSingleFileReadabilityTests.AWriteReplacesAFileExactlyWhenItReads` (Mesh), `a_file_is_replaced_exactly_when_it_reads` (RS:5132), "SingleFileMessagePackBackingStore replaces a file exactly when it reads" (TS test :2321) and `test_single_file_replaces_a_file_exactly_when_it_reads` (Python). AUTH cases: `packages/cultnet-ts/test/idunn-runtime-authority.test.ts`. Declared divergences in the manifest: `legacy-envelopes.bin` (C# refuses; the others read and replace it) and `variant-slot-v3.bin` (C# reads; the others refuse).
- **Known gaps.**`variants:follow_up:c0-bad-canonical-in-manifest` (`current-catalog-bad-canonical.msgpack` is not in the manifest; TS reads it, Rust refuses). `variants:follow_up:legacy-vectors-all-runtimes` (the three `legacy-catalog-*` c0 vectors have a verdict only in C# tests; all four runtimes refuse them, probed). `variants:follow_up:range-check-in-ci` (`npm run test:pack:ts` builds every admitted tag and checks exports, but `.github/workflows/publish-packages.yml` does not run it).
Data for the per-consumer sweep in docs/document-variants-cut.md, C2. Every object type below sits in a list (`List<T>`, `T[]`, `IEnumerable<T>`, and the like) reachable from a registered document type, and needs one string member with `[Key(n)]` and `[CultElementId]` before that consumer can take a CultLib that requires ids. A union base takes the member on the base; its subclasses inherit it, so put the slot above the subclasses' highest key.
Method: a static scan of `F:\Projects\<consumer>` C# files for types marked `[CultDocument(`, walked through keyed members, base classes, dictionary values and `[Union]` subtypes. Read-only; no consumer was edited. Worktree copies (`Aetheria-catalog`, `CultLib-*`, and so on) are excluded. Struct elements are values and need no id. Non-C# consumers carry the id as an ordinary member and are not listed. The scan is text-based: a document registered without a literal `[CultDocument(` attribute, or a type whose members lack an integer `[Key]`, is missed, so the sweep pass should confirm each repo by registering its types once.
Lines: `Type file:line (via Owner.Member)`, paths relative to the consumer root.
These types are a vendored copy of CultLib (`vendor/CultLib`); C2a already adds the ids in CultLib, so they resolve when the pin bumps. The witness bundle types are the fork reported with C2a.
`CultMeshCdnChunkRef` (`src/GameCult.Mesh/CultMeshCdn.cs:73`) has a derived id, `[CultElementId(nameof(Offset))]`: the decimal text of its offset, never random, so packing the same body encodes the same manifest bytes. `contracts/cultmesh/content-vectors.*.json` and the Rust runtime carry it. A consumer with content-addressed elements declares the same way.
Eyes pass, 2026-10-22. Facts and evidence pointers only; no recommendation.
Question as admitted: Eve.Surface and Brokkr build against an unpinned sibling `..\CultLib` source checkout. C2a (refuse element types with no id) lands on main. Options: pin-then-land, sweep-then-land, land-and-break.
| S4 | Software Engineering at Google, ch. 16 Version Control and Branch Management | https://abseil.io/resources/swe-book/html/ch16.html | fetched (does not cover live at head) |
Anything not in a "fetched" row above is not claimed from the web.
## 1. Established practice: who pays for a breaking upstream change, and when
- **Abseil / live at head (S1, S2).** Recommends users track the latest code. Says it will provide a tool to perform an API refactor for well-behaved code (S1). Requires of downstream: show how you call it, stay within the compatibility guidelines; otherwise fork or use periodic supported branches (about every 6 months, patched 2 years) (S1). Says every change is potentially breaking for someone; breaks are announced in commit messages and LTS notes, and users are asked to report undetected breaks on GitHub (S2). Users should be able to temporarily pin to an earlier commit if a newer change is a problem (S2). S1/S2 do not mention Rosie, the Beyonce rule, or caller-fixing; the fetched pages describe tooling and a feedback channel only.
- **Live at Head as described in SWE at Google ch. 21 (S3).** Providers are expected to test changes against downstream dependents before committing; breaking changes wait until dependents are updated or an automated migration tool exists. Consumers must keep meaningful tests. Stated preconditions: unit tests across dependencies run in CI, a visible reverse-dependency graph, substantial compute, no pinning of dependencies, provider acceptance of the downstream-testing duty. Called "costly but effective" inside Google and "largely unproven at scale." Notes that under SemVer a major bump costs maintainers little and many indirect users a lot. Ch. 16 (S4) does not address live at head; it states large-scale changes across trunk are costly and tooling-supported.
- **Go MVS (S5).** Selects the oldest version satisfying all declared minimums. Publishing a new version changes nobody's build; a consumer gets it only when a requirement names it, directly or through another module. Upgrades are explicit edits to `go.mod` by the top-level module. Compatibility rule: newer versions inside one major version are expected to work where older did; a major version is a different module identity. Cost of a breaking change falls on whoever opts into the new version, at the time they opt in.
- **Cargo (S6).** Git dependency without `rev` takes the default-branch tip at first build and fixes that commit in `Cargo.lock`; changes arrive only on `cargo update`. A `rev` in `Cargo.toml` pins explicitly; the Cargo book calls hand-managed SHAs "tedious and error prone" and prefers the lockfile. Consumer pays at `cargo update` time.
- **npm (S7).**`package-lock.json` records the exact tree; installs stay on it until the lockfile changes and is committed. It is not published with a package. The page gives no timeline for when updates reach a consumer.
- **Chromium DEPS (S8).**`DEPS` stores dependency URLs and commit hashes (often in `vars`). Developers roll with `gclient setdep`, `roll-dep`, or manual `git update-index`; some projects (Skia) are autorolled. The page says manual rolling after an upstream change is good practice to confirm the roll compiles and tests pass. It does not describe revert procedure. Cost is paid by the roller at roll time.
- **Git submodules (S9).** The superproject records one commit SHA (gitlink, mode 160000). Upstream commits never move the pin; it moves when someone updates the checkout and commits the new SHA in the superproject. Collaborators receive the pin via pull and must run `git submodule update`.
## 2. Internal precedent
- **Aetheria pin (L1).**`Directory.Build.props` holds `CultLibRevision` (currently `45c2f4006f5f7dab357426cabd3197e8d0a17f66`) and `CultMathRevision` (`9c1b06cd703bfa44352050e819b0d0659cf5978a`). `Directory.Build.targets` target `VerifyCultLibRevision` (BeforeTargets `ResolveProjectReferences`) errors if the sibling checkout's `git rev-parse HEAD` differs from the pin, or if `git status --porcelain` is non-empty; escape hatch `SkipCultLibRevisionCheck=true`. File is 39 lines; the pin is one pair of properties per pinned package.
- **Why added (L1 git).** Commit `4f4beb0f` (2026-09-14, "Pin CultLib for headless builds"): "Directory.Build.targets carries Delvehold's VerifyCultLibRevision guard, checking for CultMath.csproj: HEAD must equal the pin and the worktree must be clean." Commit `e91ba369` (2026-09-24, "Pin CultLib per package, as Unity's manifest does"): headless build must compile the revisions Unity consumes; a single pin would have compiled six commits of newer CultCache under headless tests while Unity stays on its tag. The Aetheria cut doc (L2, ~line 1905) calls the guard "the only working pin for a sibling checkout." Aetheria's `Packages/manifest.json` pins Unity packages by git tag.
- **Pin guard already in Delvehold (L4/search_sources hit, `Delvehold:Directory.Build.targets`)** and a variant in AetheriaEve (archived, `Aetheria.State.Dependencies.targets`, also checks a Unity package tree hash). Delvehold's guard checks `GameCult.Mesh.csproj` exists, HEAD equals `CultLibRevision`, tree clean. Date of Delvehold's first guard: not determined.
- **Consumer table (L2, Q6 in Aetheria `docs/cultcache-migration-cut.md`).** Aquarium, Ymir, AquaSynth, Mimir, Gjallar: ".NET sibling, unpinned", compile against main at once. Brokkr, Eve, EveUnity, EvePlugins: "Unity package / sibling", pin "various". Delvehold: sibling with revision guard at `334e60f`, 52 behind, "Not until it re-pins." Heimdall: submodule `b6b1d9c`. Rust consumers (Idunn, Epiphany, Odin, Ghostlight, Muninn, Ratatoskr, CodexConnector): Cargo by commit.
- **Operator discussion (L4).** voidbot `search_history` ("pin CultLib revision vs track main for consumers"; "Brokkr Eve sibling CultLib checkout build pin") returned no relevant operator statement on pinning versus tracking; hits were unrelated Discord chatter. Nearest operator words found in the Discord index: Metacrat, 2026-05-09, "Oh, anyone using CultLib will still be paying the blood price, you can count on that" (message 1502671320406687894; context not fetched, so its subject is not confirmed). No operator quote on pinning versus tracking was found in memory files grepped (`C:\Users\Meta\.claude\projects\F--Projects\memory`, pattern pin/track main/unpinned/sibling) beyond a retired Idunn pin line and Odin/Muninn pin-branch coordination notes.
- Aetheria cut doc operator mentions near pins are about content (Schematic-shape drawer, 2026-09-14; Studio grouping, 2026-09-17), not pinning policy.
## 3. Current state of the two consumers (L3)
- **Eve.**`packages/org.gamecult.eve.surface/GameCult.Eve.Surface.csproj:16``CultLibRoot` defaults to `$(MSBuildThisFileDirectory)..\..\..\CultLib`; lines 20-21 ProjectReference `$(CultLibRoot)\src\GameCult.Caching\...` and `...\GameCult.Mesh\...`. `tests/GameCult.Eve.Surface.Serialization.Tests/*.csproj:19` references `$(CultLibRoot)\src\GameCult.Mesh\...` (and Eve.Surface by relative path). `git ls-files` shows only `packages/org.gamecult.eve.surface/Directory.Build.props` (sets only BaseOutputPath/BaseIntermediateOutputPath/MSBuildProjectExtensionsPath, no revision). No `Directory.Build.targets`, no `global.json`, no `.gitmodules`, no revision property or check. Eve also holds Cargo.toml/Cargo.lock (Rust side, not examined).
- **Brokkr.**`brokkr-command/Brokkr.Command.csproj:7`, `brokkr-contracts.Tests/Brokkr.Contracts.Tests.csproj:7`, `brokkr-contracts/Brokkr.Contracts.csproj:8`: `CultLibRoot` defaults to `..\..\CultLib`. ProjectReferences: Caching, Caching.MessagePack, Mesh, Networking (by project). No `Directory.Build.*`, no `.gitmodules` in `git ls-files`; no revision property or check.
- Neither repo contains `CultLibRevision`, a `VerifyCultLibRevision` target, or any HEAD check. Both build whatever `F:\Projects\CultLib` currently holds (including dirty or other-branch states).
- Fork text claims (not re-verified here): Eve.Surface has 0 `[CultElementId]`; Brokkr `[CultDocument]` files show no object lists by grep.
Serves the C2a fork on finding variants:finding:cut-land-c2a-gate.s1.reflist-record-read-only-at-head (campaign variants), 2026-10-23.
# Prior art: read with an old writer schema, then write back
Facts only. Pages were fetched 2026-10-23 through a summarising fetcher; "not stated" means the fetched page did not say it, not that no source does. Items marked (memory) are my background knowledge, not verified from a fetched page.
- Model: the writer schema travels with the data. The reader resolves it against its own reader schema. Resolution is a read-time projection; the spec defines no write-back.
- Data loss on projection: "the writer's value for that field is ignored" when the reader schema lacks the field. A reader-only field takes its default, or the read errors if it has none.
- Write-back: the spec states no rule. Object container files store one schema in the header ("Each datum in a file is of the same schema"). `appendTo` is documented only as "Open a writer appending to an existing file"; it does not say which schema the append uses. (memory: Java `appendTo` reuses the file's header schema, so an append never restamps.)
- Stated reason for the read-time design: readers and writers evolve separately (fields added or removed in a different software version).
- On write: unknown fields are "included during parsing and in the serialized output" (matches proto2). Round-trip keeps fields the current type does not know.
- Loss cases the page lists: serializing to JSON, and copying a message field by field into a new message (it recommends `CopyFrom`/`MergeFrom`).
- No schema id or version is written. The wire format carries only field numbers, so "restamp" has no meaning here.
- Not stated: behaviour in versions before 3.5 (memory: proto3 dropped unknown fields before 3.5, then reversed). The Java reference page says nothing on preservation.
- Stated reason: preserve data across a mixed-version fleet.
- BACKWARD (default): consumers on the new schema read data written with the new or the previous version. FORWARD: data written with the new schema is readable by consumers on the old one. FULL: both.
- Write: "Each schema version gets a unique ID and an incremented version number." A producer writes under the schema it holds (its own current version). The id is stamped per message at serialisation, from the producer's schema.
- Upgrade order is how the rules prevent loss. BACKWARD means upgrade consumers before producers. FORWARD means upgrade producers first, and keep old-schema data away from consumers until they are upgraded.
- Not stated on the page: what a consume-modify-produce app writes. (memory: it re-serialises under its own latest schema, so fields it does not model are lost unless a compatibility rule kept them.)
- The registry gates registration. It does not gate individual rewrites.
- Documents carry `schema_version` (absent means 1). The app keeps a handler per version and writes new documents at the current version.
- Migration strategy is the app's choice: migrate all, migrate "when accessed" (lazy), or never.
- Not stated: whether an update of an old document restamps it. The manual's update example branches on `schemaVersion` and writes into each version's own shape. It notes its `else: null` branch adds stray null fields to the other shape.
- Stated reasons: no downtime, failed migrations easier to handle, timing flexibility. Cost: handler code per live version.
- Per-file schema version. A changed schema needs an incremented `schemaVersion`, plus a `migrationBlock` for non-trivial changes. The file is migrated whole at open, before any read or write, and Realm "will throw an error" if the version is lower or the same but the schema differs.
- No lazy per-object restamp. Writes happen only on a migrated file.
- Data loss: the manual says "Deleting a property will not impact existing objects", i.e. removal is automatic and needs no developer step. Required new properties and renames need a manual block. (memory: removed-property values are dropped from the file.)
### Core Data
- Source: web search results only (Apple forum threads 745114 and 785427, WWDC22-10120, "staged migrations" doc). The Apple docs URL I tried returned 404.
- A store whose model hash differs from the current model fails to open with an incompatible-version error. Opening migrates the whole store when `NSMigratePersistentStoresAutomaticallyOption` and `NSInferMappingModelAutomaticallyOption` are set. Lightweight migration covers only obvious patterns (add entity, remove relationship, required to optional, rename). Anything else needs a staged or custom migration.
- Migration is store-wide at open. It does not happen per record on write.
### EF Core
- Not fetched. (memory: no per-row schema stamp. Migrations are explicit, ordered DDL, tracked in a `__EFMigrationsHistory` table.)
## 5. Old-schema data held read-only until explicit migration
### Elasticsearch archive indices
- Sources: https://www.elastic.co/docs/deploy-manage/upgrade/deployment-or-cluster/reading-indices-from-older-elasticsearch-versions ; https://www.elastic.co/guide/en/elasticsearch/reference/current/archive-indices.html ; 8.0 migration notes (7.0+ indices readable, 6.x or earlier must be reindexed).
- Indices from an older major are readable through search and field-caps. "They do not support the Get API or any write APIs."
- The route to a writable copy is reindex into a new index that is "fully compatible with the current Elasticsearch version". `_source` is readable, so no data is lost on reindex.
- Stated reason: the page does not say why they are read-only. It describes archive access as "slower read-only access", with limited query capability. Some legacy mappings cannot be imported (the original is kept in `_meta`), and that is the one place the page shows a lossy translation.
- Third-party (Search Guard blog, unverified against Elastic): in 9.x a plain `index.blocks.write` does not satisfy the boot check; `index.verified_read_only` is needed.
- This is the closest match to option (b) in the fetched set: an old format, readable but not writable, left until the owner reindexes it.
## Where restamp-on-write loses data (from the sources above)
- Avro: a field present in the writer schema and absent from the reader schema is dropped at read, "ignored". Any rewrite from the resolved value would omit it.
- Protobuf: unknown fields survive a binary round trip. They are lost via JSON, or when copied field by field.
- Realm: removed properties are deleted by the automatic migration.
- MongoDB: a handler that writes only the current shape can lose or relocate old-shape fields (the manual's example adds nulls).
- Confluent: loss is prevented by compatibility rules at registration time, not at each write.
## Pattern across the set
- No fetched system restamps a single record on write when the record's writer schema differs from the application's type, on an app-visible, per-record basis.
- Avro, protobuf and Confluent carry the schema per file or message.
- Realm, Core Data and Elasticsearch move or block the whole store or index at open.
- MongoDB leaves it to the app.
- Whole-store migration at open is the usual explicit step (Realm, Core Data). Elasticsearch is the one case of read-only-until-reindex.
- Not found: a system that declares per-schema-id compatibility, as in option (b). Avro schema resolution and Confluent compatibility rules are the nearest analogues. Both are read-side or registration-side declarations.
Facts and pointers only. Each source was fetched or searched this session; what was not reachable is marked.
## 1. Goetz et al., Java Concurrency in Practice (Addison-Wesley, 2006), ch. 10 "Avoiding Liveness Hazards"
Status: EXISTS, SECTION NUMBER CORRECTED.
- "Open Calls" is **10.1.4**, not 10.1.3. 10.1.3 is "Deadlocks Between Cooperating Objects"; 10.1.2 is "Dynamic Lock Order Deadlocks"; 10.1.5 is "Resource Deadlocks". Source: text extracted from a PDF copy of the book (its TOC lists Listing 10.6 "Using Open Calls to Avoiding Deadlock Between Cooperating Objects", p. 133). The copy is an unofficial one at library.megu.edu.ua; a publisher link was not checked.
- (a) 10.1.3 gives the warning sign: an "alien method" (defined p. 40 as a method whose behavior the calling class does not fully specify) called while holding a lock. Paraphrase: such a call may take other locks (deadlock risk) or block a long time, stalling threads that need the lock you hold. 10.1.4 defines an open call as a call made with no locks held (credited to Lea, CPJ 2.4.1.3) and says programs restricted to open calls are far easier to analyze for deadlock-freedom than ones that allow alien calls under lock.
- Cost it names: moving a call out of a synchronized block makes an atomic operation non-atomic (its getImage example stops being a snapshot). Where atomicity is needed it suggests structuring so only one thread can run the code path after the open call (service shutdown example).
- (b), (c): not addressed in this section.
- Pointer: JCIP 10.1.3 and 10.1.4, pp. ~132-134 (chapter 10 starts p. 128).
## 2. Sutter, "Effective Concurrency: Avoid Calling Unknown Code While Inside a Critical Section"
Status: VERIFIED, with date caveat.
- Dr. Dobb's column in the "Effective Concurrency" series. Sutter's blog teaser is dated 2007-11-06: https://herbsutter.com/2007/11/06/effective-concurrency-avoid-calling-unknown-code-while-inside-a-critical-section/ . Original link ddj.com/architect/202802983; Dr. Dobb's mirror https://drdobbs.com/cpp/avoid-calling-unknown-code-while-inside/202802983 (4 pages). The Dr. Dobb's issue date was not shown in anything fetched; the blog date is the only date confirmed. www.drdobbs.com did not resolve from this machine, so the column body comes from search-result summaries and the blog teaser, not a full first-hand read.
- (a) Core claim: locks do not compose; two separately correct modules can deadlock when combined. Trigger: caller inside a critical section, callee enters another critical section or blocks, another thread can take the same two in the opposite order. "Unknown code" listed as: library functions, plug-ins, callbacks/function pointers, virtual methods, generic types.
- Mitigations listed: move the call outside the critical section; pass copies instead of shared data; narrow the critical section (finer locks, reader-writer); require callees not to block (documentation/discipline only).
- (b) Lockdep tracks lock classes, not instances, and records dependencies between classes as sequences are observed. It reports a circular order as a potential deadlock even if the deadlock never happened. Its "Proof of 100% correctness" section says that once each simple single-task locking sequence has occurred at least once, no timing of them can deadlock. Each new lock chain is validated once and hashed; repeats skip validation.
- Nuance: it flags the dependency edge that closes a cycle, so both orders must have been exercised at some point (possibly by different, non-concurrent runs). A lone nesting is flagged only if it breaks a single-lock rule (IRQ-safe vs unsafe, same-class recursion). "Refuses on first sight" fits the closing edge or a single-lock rule violation only. It warns; it does not block the acquisition.
## 4. ConcurrentHashMap re-check after locking
Status: VERIFIED.
- Source: OpenJDK java/util/concurrent/ConcurrentHashMap.java, `putVal` and `computeIfAbsent`: head node `f` is read unlocked, `synchronized (f)` is taken, then `tabAt(tab, i) == f` is re-checked; on failure nothing is done and the loop retries. The overview comment gives the reason: the locked node can be replaced, so first-in-bin must be confirmed after locking. (Only the first 100k of 269k chars were read; both methods were reported found there.) Empty-bin `computeIfAbsent` uses a ReservationNode plus CAS instead.
- (c) This is the pattern: choose the lock from an unlocked read, validate the read after acquiring.
## 5. LMDB
Status: VERIFIED (paraphrase). Not about alien code or lock order.
- Source: lmdb.h, `mdb_txn_begin` and `mdb_env_open` (https://github.com/LMDB/lmdb, libraries/liblmdb/lmdb.h). A thread may hold one transaction at a time (plus children); a transaction and its cursors belong to one thread; one write transaction at a time, so writers cannot deadlock each other; MDB_NOTLS ties reader slots to the txn object instead of the thread, and the application must then serialize writers itself because write locking is unaware of user threads.
- Relevance: "one hold per thread" as a deliberate rule, parallel to CultCache's one-cache-gate-per-thread.
## 6. SQLite btmutex.c
Status: VERIFIED (summary of trunk source).
- Source: https://github.com/sqlite/sqlite/blob/master/src/btmutex.c (read via raw.githubusercontent.com; sqlite.org served a bot check).
- (b) Prevention by ordering, not detection: sharable B-trees on a connection sit in a list sorted by BtShared address and are always locked ascending (`sqlite3BtreeEnterAll`). `sqlite3BtreeEnter` try-locks first; on failure it releases later-ordered locks, waits, then reacquires in order. `wantToLock` is a count that makes a non-recursive mutex behave recursively.
- No alien-code or callback discussion in these comments.
## Counter-positions and how they stay safe
- Callbacks under lock, documented: Java `ConcurrentHashMap.compute`/`computeIfAbsent`. Javadoc (Java 21): the function should be short and simple because other threads' updates may block meanwhile, and must not modify the map. The javadoc does not say "internal lock"; that it runs under the bin lock is inferred from the source in item 4. Safety rests on documented restrictions, not enforcement.
- Opposite choice: .NET `ConcurrentDictionary.GetOrAdd(key, valueFactory)`. Writes use fine-grained locks but valueFactory runs outside them, "to avoid the problems that can arise from executing unknown code under a lock" (doc remarks). Price: not atomic; the factory may run more than once; a recheck decides the winner (another thread's insert wins). https://learn.microsoft.com/en-us/dotnet/api/system.collections.concurrent.concurrentdictionary-2.getoradd . Same shape as option `encode-first` (user code outside the lock, then recheck and commit) and as item 4.
- JCIP and Sutter both name the cost of leaving the lock: lost atomicity. Sutter also offers "pass copies" and "callee must not block" as alternatives to leaving.
- Not researched (no source fetched): Linux notifier chains, Java `synchronized` event listeners, other systems that call back under lock by design. No verified citation for them.
## Unverified / not done
- JCIP: section numbers and text come from one PDF copy, not the publisher edition.
# Prior art: a process-global default registry deciding write ownership
Serves `variants:question:registryless-write-reads-shared` (campaign `variants`). Eyes pass, session `self-2026-10-27b-eureka`, 2026-10-27. Facts only; items marked "summary" rest on search summaries not opened.
# Prior art: process-global type registries and write/overwrite ownership
Fork: `variants:question:registryless-write-reads-shared` (viewed in full via eureka-state, status InForce, options `shared-decides` / `written-type-only`, recommended `shared-decides` at ~65%).
Facts only. No conclusion. Date 2026-10-27.
Method note: web pages were read through WebFetch (a small model summarises the page) and WebSearch summaries. Quotes below are what those tools returned; where only a summary was available it is marked "summary". Nothing was run.
## A. Local mechanics (CultLib, read-only, `C:\cl-wt\c2a-land\src\GameCult.Caching\CultCache.cs`)
- `CultDocumentRegistry.Shared` is `Lazy<CultDocumentRegistry>` (line 639-645). Its public constructor calls `Refresh()` (647-650), which scans `ReflectionExtensions.GetAttributedDocumentTypes()` (688; defined in `ReflectionExtensions.cs:10`) and rebuilds the indexes under `_registrationGate`. So Shared's initial content is whatever `[CultDocument]` types are in loaded assemblies at first touch of `Shared`.
- `GetRequired(Type)` (697-711) registers an unregistered `[CultDocument]` type on demand. So Shared also grows whenever any caller asks for a type. `Refresh()` is public and rebuilds the index from currently loaded assemblies.
- `ForTypes(IEnumerable<Type>)` (658-669) builds a registry with `discoverLoadedDocuments:false`, containing only the given types (plus those `GetRequired` pulls in).
- `WriteDocument<TDocument>(key, doc, storedAt, registry = null)` (4515-4545): `resolvedRegistry = registry ?? Shared`; `catalogEntry = resolvedRegistry.GetRequired<TDocument>().ToCatalogEntry()`; then `Cacheless(resolvedRegistry, ...)` lends that registry to the store for the call. XML doc (4510) says Shared "can only refuse more".
- `WriteRecord(..., registry = null)` (4491-4497): XML doc (4489) says that with registry omitted "it decides nothing for copied records and the one entry is the only claimant".
- `OwnsOnDisk(upserts)` (4741-4759): when the store is cacheless and upserts are given, ownership of an on-disk record is decided by `CultDocumentRegistry.Owns(id, catalog, declared, new CatalogEntryLookup(writes, AttachedRegistry), ..., claimant => writes.Any(entry => entry.SchemaId == claimant.SchemaId))`. The comment (4731-4735, 4738-4740) says written entries shadow the lent registry's type for their ids, the registry adds other claimants, and the record is owned only when the claimant is a written entry, "a registry can only refuse more". `CatalogEntryLookup` ctor at 604 takes `(entries, CultDocumentRegistry? others)`.
- Net: with `registry == null` the lent registry's claimant set is Shared's current contents, which depend on assemblies loaded and on every prior `GetRequired` call in the process.
## B. Q1. Process-global, lazily filled registries in established libraries
### MongoDB C# driver, `BsonClassMap`
- Global: yes. `BsonClassMap.RegisterClassMap` is process-wide static; unregistered types are auto-mapped on first use. Docs: "You must register a class map _before_ it's needed in your code." and "We recommend registering class maps before initializing a connection with MongoDB." https://www.mongodb.com/docs/drivers/csharp/current/serialization/class-mapping/ (WebFetch; the page itself says nothing on freezing or late-registration errors).
- Behaviour depends on what has been touched: EF Core provider issue EF-88 reports "Unknown discriminator value 'List`1'" when the first operation after startup is a read; reporter links it to a class map that is created only on write. Status Closed/Fixed in 8.0.0; page has no maintainer comment. https://jira.mongodb.org/browse/EF-88 (WebFetch). Reporter says explicit `RegisterClassMap` at startup did not fix it for them.
- A MongoDB community answer (via search summary, not opened): discriminator mappings persist once registered, so a read after an insert works; fix offered is `[BsonKnownTypes]`. https://mongodb.com/community/forums/t/c-driver-why-is-bsonknowntypes-attribute-needed-for-polymorphism/2183 (summary only).
- Register-once contract: CSHARP-4424 states the three `RegisterClassMap` methods are "meant to only be called once for a given type"; requested because several libraries share data models and do not control init order. Closed Done, fix version 2.19.0; the page does not say which API shipped. https://jira.mongodb.org/browse/CSHARP-4424 (WebFetch). Community workaround named in search summary: `BsonClassMap.IsClassMapRegistered`.
- CSHARP-1879 (search summary): `RegisterClassMap` accepted an unfrozen map and failed later on use; fixed 2.4.2. https://jira.mongodb.com/browse/CSHARP-1879 (not opened).
- CSHARP-5597 (search summary): `BsonSerializer.RegisterDiscriminator` has no effect; discriminator dictionary populated at class-map registration. https://jira.mongodb.org/browse/CSHARP-5597 (not opened).
- Polymorphic docs: "You must specify hierarchical discriminators in root-to-leaf order." and the relationship of all classes in a hierarchy must be documented before deserialization. https://www.mongodb.com/docs/drivers/csharp/current/serialization/polymorphic-objects/ (WebFetch).
- Particular docs report an order-dependent GuidSerializer failure ("incorrect order of class mappings"). https://docs.particular.net/persistence/upgrades/mongodb-4to5 (search summary, not opened).
- Did they move away: not found. Class maps remain global static; mitigation is documentation plus the IfNotRegistered API.
### protobuf-net `RuntimeTypeModel.Default`
- Global: yes. README: "The Serializer.* methods are basically just shortcuts to RuntimeTypeModel.Default.*". https://cdn.jsdelivr.net/gh/protobuf-net/protobuf-net@main/README.md (WebFetch).
- Lazy fill, freezing, order-dependent bugs: NOT verified. README is silent; a direct docs fetch (docs.protobuf-net.dev/runtime) was redirected and not followed. No issue found.
### MessagePack-CSharp
- Global mutable default exists: `MessagePackSerializer.DefaultOptions`. README (WebFetch): labelled "Global state; Not recommended when writing library code." Safe if you control all MessagePack code; libraries should pass options explicitly because others "may change this static property." Libraries "must not use StaticCompositeResolver - as it is global state". https://github.com/MessagePack-CSharp/MessagePack-CSharp/blob/master/README.md
- Resolvers are an ordered list (`CompositeResolver`): order matters by construction.
- Whether it moved away from the global: not found.
### System.Text.Json
- `JsonSerializerOptions.Default`: "Gets a read-only, singleton instance"; remarks give a performance reason: shared instance avoids "redundant recomputation of converters". https://learn.microsoft.com/dotnet/api/system.text.json.jsonserializeroptions.default (WebFetch). The page states no rationale about cross-component interference; any such reading is inference. Design-discussion issue in dotnet/runtime: searched, not found. Whether Default was ever mutable: not verified.
- Type-info resolvers: not researched beyond the above.
### Json.NET
- `TypeNameHandling` is a security concern, not an order concern: CA2326 says "An attacker could modify the serialized data to include unexpected types"; mitigation is a custom `ISerializationBinder` restricting types. https://learn.microsoft.com/en-us/dotnet/fundamentals/code-analysis/quality-rules/ca2326 (WebFetch).
- `JsonConvert.DefaultSettings` global behaviour/guidance: the Newtonsoft page fetched says nothing about global effects. Not verified.
### Orleans
- Serializer discovery is build-time source generation (`[GenerateSerializer]`, `[Id]`), not a runtime scan; the docs' older code-generation section lists scan of assemblies referencing the core library plus `KnownType`/`KnownAssembly` attributes. Aliases: "Type aliases are globally scoped, and you cannot have two aliases with the same value in an application." https://learn.microsoft.com/en-us/dotnet/orleans/host/configuration-guide/serialization (saved locally: `C:\Users\Meta\.claude\projects\F--Projects-CultLib\0840c7c7-0e31-410e-9012-afb79c3c0742\tool-results\toolu_0148m7no1AbrJzWmNLicn6N9.txt`, lines 134, 839-841).
- Type manifest / codec registry runtime behaviour, issues: not researched.
### EF Core
- Model is built per `DbContext` type and cached (IMemoryCache; "the built model has a cache size of 100"); initialised at first operation: "just creating a DbContext instance does not cause the EF model to be initialized." https://learn.microsoft.com/en-us/ef/core/performance/advanced-performance-topics
- Assembly scan: `ApplyConfigurationsFromAssembly` — "The order in which the configurations will be applied is undefined, therefore this method should only be used when the order doesn't matter." https://learn.microsoft.com/en-us/ef/core/modeling/ ; and "`EntityTypeConfigurationAttribute` types will not be automatically discovered in an assembly."
- So EF's registry is per context type, not process-global; assembly scan is opt-in and documented as order-undefined.
### Marten
- Storage is a table and upsert function per document type, alias `mt_doc_<alias>`; alias collisions are resolved with `DocumentAlias`. https://martendb.io/schema/storage (WebFetch). Hierarchies share one table via `AddSubClass<T>()`; subclass docs are "stored as the parent type" in the identity map. https://www.martendb.io/documents/hierarchies (WebFetch). Schema is configured in `StoreOptions` per store, not a process-global.
### Java ServiceLoader
- "Service providers are loaded and instantiated lazily, that is, on demand." Providers on the class path load in the order `getResources` finds config files; "The ordering of modules in a module layer is not defined." "Instances of this class are not safe for use by multiple concurrent threads." Caches cleared by `reload`. https://docs.oracle.com/en/java/javase/21/docs/api/java.base/java/util/ServiceLoader.html (WebFetch). Visibility is class-loader dependent ("must be visible to the class loader").
## C. Q2. Write/overwrite decisions against an existing record of another type
- RavenDB: REST put docs, request body section: updates must "include its collection name in the metadata" and "A document's collection cannot be modified." Server error code for mismatch: the first fetch summary said HTTP 500; not confirmed on a second fetch. https://docs.ravendb.net/7.1/client-api/rest-api/document-commands/put-documents (WebFetch twice). Ayende, RavenDB 4.0 storage: document key namespace is global, same key cannot exist in two collections (summary). https://ayende.com/blog/173954/the-design-of-ravendb-4-0-physically-segregating-collections The check is against the collection named in the write (derived from the writer's type by the client), not a registry; the "client maps entity type to collection" part is inference from the search summary, not a quoted source. `PutDocumentCommand` page: a missing collection means `@empty`; concurrency via `changeVector`, null "to skip the check". https://docs.ravendb.net/client-api/commands/documents/put
- Marten: per-type tables, so the same id under two types are two rows in two tables; no cross-type check. Same id across subclasses of one hierarchy table: not covered by docs fetched; unverified. Marten docs: upsert is used so users "don't have to check whether a document is new or existing" (search summary of https://www.martendb.io/documents/storing, not opened).
- MongoDB: the server has no type; `_id` unique per collection; the C# driver writes a `_t` discriminator from the written object's class map. Whether the driver checks an existing document's `_t` on replace: not researched; nothing found stating it does.
- Couchbase: key unique within a collection; upsert replaces; no server-side type field enforcement; guidance is one type per collection. https://docs.couchbase.com/server/7.0/learn/data/scopes-and-collections.html (search summary, not opened).
- Orleans: grain identity is type plus key ("type/key"), so same key under two grain classes is two identities, not a conflict. https://learn.microsoft.com/en-ca/%20dotnet/orleans/grains/grain-identity (search summary, not opened). How storage providers key state: not found.
- Result of this pass: no store found in which a process-global type registry decides whether a write may replace an existing record of another type. Found: collection/type fixed by metadata of the writer (RavenDB), type as part of the key (Orleans, Marten tables), or no check (Couchbase upsert). This is a negative search result, not proof of absence. Not searched: Cosmos DB, DynamoDB, EF Core TPH update of an existing row with another discriminator, Realm, LiteDB, Akka.Persistence.
## D. Q3. Guidance on ambient/global registries and determinism
- Mark Seemann (ploeh, 2019): after discussion with co-author van Deursen, "Ambient Context really is an anti-pattern." Reasons given: he has never used it and superior alternatives exist; he says the claim is unproven. https://blog.ploeh.dk/2019/01/21/some-thoughts-on-anti-patterns/ (WebFetch). Examples he cites as found in the BCL (search summary, not verified on page): `Thread.CurrentPrincipal`, `CultureInfo.CurrentCulture`, `HttpContext.Current`. DI book 2nd ed. chapter 5 lists it among DI anti-patterns: https://www.oreilly.com/library/view/dependency-injection-principles/9781617294730/OEBPS/Text/c05.xhtml (not opened).
- MessagePack README (above) is itself a library-author statement on global mutable defaults. MongoDB CSHARP-4424 (above) states the shared-library init-order problem. EF Core docs (above) state assembly-scan order is undefined.
- No .NET guidance specifically about AppDomain-wide/assembly-scan registries and write decisions found.
## E. Not verified / gaps
- protobuf-net lazy registration, freezing and bug history.
- Why System.Text.Json made Default immutable; no maintainer statement found.
- Json.NET DefaultSettings guidance.
- RavenDB exact error text and HTTP status; whether the stored document is unchanged after refusal.
- Marten same-id-across-subclasses behaviour.
- MongoDB driver behaviour on replacing a document whose `_t` differs.
- Several items above rest on WebSearch summaries I did not open (marked "summary").
# Prior art: structural versus name-inclusive schema ids
Serves the store-identity question raised 2026-10-27 in campaign `variants` (operator: member and declared names vary with runtime naming conventions). Eyes pass, session `self-2026-10-27b-eureka`. Facts only; items marked "summary" rest on search summaries not opened.
# Prior art: structural vs name-inclusive schema identity (campaign variants)
Eyes pass, 2026-10-27. Facts and evidence pointers only. "(fetched)" = page opened via WebFetch (the answer is a small-model extraction of the page, not raw text). "(summary)" = WebSearch result summary, page not opened.
## 0. Local facts (read-only)
- `F:\Projects\CultLib\src\GameCult.Caching\CultCache.cs:717-718` schemaId = Sha256(BuildSemanticFingerprint(...)). `:735-761` fingerprint = `schemaName|schemaVersion` then per member ordered by Slot: `slot:MemberName:TypeName:ref|value:TargetSchemaName:one|many`. `contentHash` (`:716`) is a separate Sha256 of canonical JSON (`:763+`).
- `TypeName` = `CultSchemaTypeNames.FromType`: CLR `Type.FullName` (namespace-qualified), generics as `Name<args>`, arrays `T[]` (`src\GameCult.Caching\CultSchemaTypeNames.cs:9-31`). So the fingerprint contains .NET type names as well as member names.
- Contract `Contracts\cultcache-persistence-format.md:257-284` lists what canonicalization must preserve: slots, types, required/default, nullability, collection semantics, enum constraints, encoding shape. Member names and declared schema name are not in that list. `:171-179` says the id is not derived from C#/Rust/TS type names and the same shape should give the same id "regardless of origin language". `:180-190` defines schemaId = "same semantic persistence shape?" vs contentHash = "exact same canonical schema document?". Example catalog entry (`:140-170`) carries optional `supersedesSchemaIds`, `migrationFamily`; its example canonicalSchema fields do carry `"name": "Name"` (PascalCase). Soft-migration reader flow `:286-345` (exact_match / soft_migrate / reject).
- Runtimes do not derive ids; they emulate C# names:
- Python: `packages\cultcache-py\src\cultcache_py\documents.py:326``member_name=_pascal_case(field.name)`; `:516-517` splits on `_` and capitalises. `schema_id` defaults to `self.type` (`:218`); `compatible_schema_ids` field (`:188,242`).
- TypeScript: `packages\cultcache-ts\src\codec.ts:226-232``pascalCase` (splits on non-alphanumerics), `:267``memberName: pascalCase(field.name)`; throws if a codec has no C# type name (`:~262`).
- Rust: `packages\cultcache-rs\src\lib.rs:295,308``serde(rename_all = "camelCase")` on envelope structs; `schema_id: Some(T::TYPE.to_string())` at `:2856`. I did not trace how Rust document types obtain a sha256 id.
- Python test pins a literal id from shared vectors: `packages\cultcache-py\tests\test_cultcache.py:808` (`tests\vectors\document-variants-c0`).
- C# also has `CompatibleSchemaIds` (`CultCache.cs:28,313,570`): a record whose persisted id is in the list is accepted (an alias-list mechanism; I saw only the grep lines, not the logic).
- Ruling `variants:ruling:wire-id-is-version` (eureka-state, InForce, ruled 2026-09-30, operator quote "Agreed on schema identity"): wire schema id = declared version string `<name>.vN`; every runtime stamps/resolves it exactly; a wire binding without `.v<digits>` is refused; "Store ids are unchanged." Cited source: document-variants-cut.md:552-555 (not opened).
## 1. How established formats define identity/compatibility, and whether names participate
Protocol Buffers (fetched https://protobuf.dev/programming-guides/proto3/ and /json/)
- Field number identifies the field in the binary wire; changing it = delete + add. Field names appear only in TextProto/JSON encodings; reserve deleted numbers and, for JSON/text, names. The proto3 page has no explicit safe/unsafe entry for renaming; names matter only for JSON/text.
- JSON mapping: names become lowerCamelCase unless `json_name` is set; "Parsers accept both the lowerCamelCase name" and the original proto name (/programming-guides/json/).
- Message type identity in `Any` is by name: default type URL `type.googleapis.com/<package>.<message>`, "a globally unique identifier" (proto3 page, #any). Type identity is nominal; compatibility is by field numbers.
- Parsing Canonical Form (PCF): STRIP keeps only attributes relevant to parsing (type, name, fields, symbols, items, values, size); doc and aliases dropped; FULLNAMES replaces short names with fullnames and drops namespace attributes. Textually equal PCF means "the same" for any reader. Names ARE in the canonical form (record and field names), as is field declaration order. Defaults are absent from the kept list (extractor's inference; spec text not confirmed).
- Fingerprints: CRC-64-AVRO in single-object encoding (marker C3 01 + 8-byte LE fingerprint); spec recommends at least a 64-bit Rabin, with SHA-256 / MD5 options for larger sets.
- Name rules: `[A-Za-z_][A-Za-z0-9_]*`, dot-separated namespaces, equality on case-sensitive fullname.
- Resolution: records match on same unqualified name; "Named types and fields may have aliases"; implementations "may optionally" use aliases to map writer to reader. So in Avro names participate in identity (fingerprint) and resolution; aliases are the rename mechanism and are excluded from the fingerprint.
Cap'n Proto (fetched https://capnproto.org/language.html)
- Files and types carry 64-bit IDs; `capnp id` generates them randomly. Default ID for a declaration without one = first 8 bytes of MD5 of (parent scope ID, declaration name). "You cannot change the name of a type that doesn't have an explicit ID"; symbolic names can change if ID/ordinals stay. Stated rationale for IDs: short, unambiguous, full names waste wire space. The page does not say why structural hashes were rejected; I found no Varda statement (search pointed at a mailing-list thread "Schema Refactoring and Unique IDs", Aug 2023, unopened).
- Encoding page (https://capnproto.org/encoding.html, fetched): adding fields is backward compatible because field positions depend only on lower-numbered fields; nothing on type identity.
FlatBuffers (fetched https://flatbuffers.dev/evolution/ and /schema/)
- Field positions by declaration order or explicit `id: n` (contiguous from 0, on all fields if used). Renaming "generally OK" because names are not in the binary. `file_identifier` is exactly 4 chars at bytes 4-7; docs frame it as a file-level check and recommend unions for distinguishing messages.
- Binary protocol: field header = type + field id; "the field name is not encoded". (IDL page gave nothing on omitted ids.)
MessagePack-CSharp (fetched https://github.com/MessagePack-CSharp/MessagePack-CSharp, first 100k of 105k chars)
- `[Key(int)]` on all members => array encoding; string keys => map. Keys should be sequential from 0 (gaps cost nulls); do not reuse indexes of removed members; keep obsolete members until clients update. Union keys are ints, unique, serialized as [key, object]. No type identity in the payload.
Microsoft Bond (fetched https://microsoft.github.io/bond/manual/bond_cpp.html)
- Changing field ordinals breaks wire compat; no explicit versioning; Simple JSON matches by field name so a rename breaks it. No schema hash/GUID documented there. Bebop: wiki page empty (points to docs.bebop.sh, not opened).
Unison (fetched https://www.unison-lang.org/docs/the-big-idea/ and /docs/language-reference/unique-types; hash size from search summary)
- Definitions identified by hash of syntax tree; names are "separately stored metadata"; hash excludes names (512-bit SHA3 per search summary, unopened).
- Structural types: "Unison does not uniquely identify a type by its name"; same-shape types are the same type (Suit/Direction, four nullary constructors, interchangeable). `unique` types: compiler generates a UUID that feeds the hash, so identity becomes nominal. Which is the default keyword was not confirmed.
IPLD (fetched https://ipld.io/docs/schemas/)
- Matching is structural: "the names of types aren't expected to be found in the serialized data"; match is by structural outline. The page does not say schemas are themselves content-addressed. Versioning is left to developers; "Migrations" subpage not opened.
## 2. Failure modes of purely structural identity
- Unison documents it directly: structurally identical enums are one type unless declared `unique` (unison-lang.org/docs/language-reference/unique-types, fetched). Its mitigation is a per-declaration UUID mixed into the hash (identity re-nominalised).
- Language literature (summary, pages not opened): structural typing is prone to accidental conformance/type equivalence; nominal typing prevents it at the cost of rigidity (Wikipedia "Nominal type system"; arXiv 1007.2123 lecture notes; Erlang nominals doc https://www.erlang.org/doc/system/nominals.html says nominal types exist to prevent misuse of same-structure types). TypeScript "brand" = phantom discriminator field. No serialization-specific source found.
- Formats that avoid the collision by carrying a name or tag outside the shape: protobuf Any type URL (full name), Avro fullname in PCF, Cap'n Proto explicit IDs, Unison `unique`, MessagePack-CSharp union key ints, FlatBuffers file_identifier/unions.
- Not found: a documented incident of two same-shape serialized types colliding under a structural id (one search, nothing).
## 3. Failure modes of name-inclusive identity across languages
- Protobuf: default camelCase conversion is "always valid but not necessarily unique"; proto3 compiler rejects JSON-name conflicts; serialized duplicate keys parse differently by runtime (Edition Zero design doc, summary: https://flutter.googlesource.com/third_party/protobuf/+/HEAD/docs/design/editions/edition-zero-json-handling.md). Snake_case vs camelCase JSON mismatches in practice: https://github.com/fullstorydev/grpcurl/issues/105 (summary); https://github.com/open-telemetry/opentelemetry-proto/issues/476 (summary: spec contradicted itself on camelCase vs original names). Protobuf's mitigation: parsers accept both forms.
- Avro restricts names to `[A-Za-z_][A-Za-z0-9_]*` with case-sensitive fullname equality, so snake_case and PascalCase spellings are distinct names; renames go through aliases, which PCF drops.
- Cap'n Proto: auto-derived IDs depend on the declaration name, so a rename without an explicit ID changes the ID.
- Bond Simple JSON: rename breaks compatibility (name matching).
- Local: Python/TS already convert to PascalCase to reproduce C#'s fingerprint (section 0); the fingerprint's TypeName is the CLR full name (`CultSchemaTypeNames.cs:30`) and TS throws when a codec lacks a C# type name.
- Registry normalisation (summary): Confluent Schema Registry `normalize.schemas=true` maps semantically-equal schemas to one id (https://www.confluent.io/blog/best-practices-for-confluent-schema-registry/); sources disagree on whether JSON Schema property order is normalised.
## 4. Migration precedents for changing identity derivation
- Git SHA-1 -> SHA-256 (fetched https://git-scm.com/docs/hash-function-transition): per-repo format marker (`extensions.objectFormat`, `compatObjectFormat`, `repositoryFormatVersion=1` makes old git refuse); bidirectional SHA-256<->SHA-1 mapping (pack idx v3, `loose-object-idx`); four transition modes (dark launch, early, late, post); conversion at the network boundary. It is a design doc, not a statement of what has shipped.
- IPFS CIDv0 -> CIDv1 (fetched https://docs.ipfs.tech/concepts/content-addressing/): v0 (46 chars, "Qm") and v1 (self-describing version/codec/multihash) coexist; `ipfs cid format -v 1` converts; v1->v0 only for dag-pb.
- Docker manifest schema 1 -> 2 (summary, mirror pages unopened): digests differ per manifest format; pull-by-digest gets no translation, pull-by-tag may be translated; repush converts. https://pulp.plan.io/issues/2099.
- Warpcast API hash v1 -> v2 (summary, https://warpcast.notion.site/Warpcast-API-Hash-Migration-Public-ccdd6de8ac604783b9596cfd623ddbac): two-week dual-hash period, then only v2.
- Cap'n Proto explicit-ID escape hatch for renamed/moved declarations; Avro aliases as the rename bridge in resolution; MessagePack-CSharp keep-obsolete-members (all fetched, above).
- Local: contract `supersedesSchemaIds` / `migrationFamily` (`cultcache-persistence-format.md:~140-170`) and C#/Py`CompatibleSchemaIds` exist as alias lists.
- Not found: Unison changing its hash derivation (one search, no docs); any serialization store that switched from name-inclusive to structural ids.
## Could not verify / not done
- Cap'n Proto designer's rationale against structural hashing (not in docs; mailing list unopened).
- Avro: whether PCF drops defaults (spec text) and any JIRA on it; spec page was summarised by the fetch model.
Element ids: a store that can hold them declares `cultcache.store.v3`; TypeScript,
Python and Rust read it and keep the header, and readers older than ids refuse it
(vectors: `tests/vectors/document-variants-c2a`).
Document variants: C# resolves them (`cultcache.store.v2`); every other runtime,
CultMesh single-file reads, the directory store and CultNet refuse a variant by name
(vector written by C#:`tests/vectors/document-variants-c1`).
@ -41,8 +44,8 @@ to speak CultNet or participate in CultMesh.
| C# | Reference implementation and production game/server substrate | CultCache typed documents, CultNet schema-v0, LiteNetLib production path with adapter profile, shared transport profiles, shared RUDP socket transport, C# multi-peer RUDP listener and schema service host, C# interop RUDP serve/dial path, connected RUDP CultMesh client helpers, RUDP-backed schema-client selection for `rudp://` service endpoints, RUDP reconnect policy/controller and reconnect-loop helper, LiteNetLib client reconnect-controller adoption, schema-client transport port for service request/response clients, CultMesh node/session/catalog/stream/simulation/RUDP surfaces, authorized peer lookup | No need to abandon LiteNetLib before the CultNet adapter fully owns production service adoption |
| Python | Package runtime, local daemon endpoint, raw interop peer, typed state participant | CultCache/CultNet/CultMesh package surface, branded CultMesh node/schema/shard/Verse/peer/authority/stream/RUDP ergonomics, authorized peer lookup, shard catalogs/logs, schema-transport factory for raw client request/subscription flows, local TCP and RUDP CultMesh server listeners, Python interop RUDP serve/dial path, session/prediction/simulation fact helpers, RUDP codec/session/socket transport, RUDP reconnect policy/controller and reconnect-loop helper, schema-v0 MessagePack over RUDP evidence, benchmark evidence against C# cache paths | Universal C# throughput, full C# daemon/server stack, and RUDP as the only public service pipe |
| TypeScript | Node/browser-adjacent projection and interop runtime | CultCache parity tests, CultNet schema-v0 peer, advertised schema/shard catalog wire contracts plus request/sync helpers, transport profiles, TCP-framed transport port and peer helper, shared RUDP socket transport, RUDP interop serve/dial path, RUDP reconnect policy/controller, RUDP reconnect-loop helper, branded CultMesh node/schema/shard/Verse/peer/authority/stream/RUDP ergonomics, authorized peer lookup | Full C# game session/server stack |
| Rust | Low-level cache/transport/runtime substrate | CultCache/CultNet interop lane, schema and shard catalog helpers, typed same-schema alias hydration from raw snapshots, transport profiles, TCP-framed port, shared RUDP codec/session/socket transport, Rust interop RUDP serve/dial path, connected RUDP CultMesh client helpers, RUDP reconnect policy/controller and reconnect-loop helper, schema-v0 MessagePack over RUDP evidence, small CultMesh facade for schema registry, shard catalog, peer cards, authority leases, authorized peer lookup, RUDP endpoint parsing, and peer-card/authorized-peer RUDP client construction | Full CultMesh game-session/server facade or production game-session server. `packages/cultmesh-rs` now holds a Rust stream and session surface -- node, document sets, RUDP document publication, stream descriptors, negotiation, stream catalog, and shared-memory frame rings -- but no interop lane exercises it, so it carries no cross-runtime parity claim |
| Kotlin | JVM/Android client substrate | Typed MessagePack cache, one-call typed snapshot document sync through local same-schema aliases, built-in schema catalog advertisement, schema/shard catalogs with shared schema-message transport fetch/sync helpers, authority leases, authorized peer lookup, stream catalog, branded `CultMesh` facade, WebSocket transport profile/connection helper, RUDP client/server sugar, connected RUDP CultMesh client helpers, Kotlin interop RUDP serve/dial path, RUDP reconnect policy/controller and reconnect-loop helper, schema-v0 MessagePack over RUDP evidence | C# SoA store (parked at `parked/cultcache-soa`), full C# server daemon, or Unity hot-loop memory layout |
| Rust | Low-level cache/transport/runtime substrate | CultCache/CultNet interop lane, schema and shard catalog helpers, typed hydration of raw snapshots into the receiver's own registered type for each schema id, transport profiles, TCP-framed port, shared RUDP codec/session/socket transport, Rust interop RUDP serve/dial path, connected RUDP CultMesh client helpers, RUDP reconnect policy/controller and reconnect-loop helper, schema-v0 MessagePack over RUDP evidence, small CultMesh facade for schema registry, shard catalog, peer cards, authority leases, authorized peer lookup, RUDP endpoint parsing, and peer-card/authorized-peer RUDP client construction | Full CultMesh game-session/server facade or production game-session server. `packages/cultmesh-rs` now holds a Rust stream and session surface -- node, document sets, RUDP document publication, stream descriptors, negotiation, stream catalog, and shared-memory frame rings -- but no interop lane exercises it, so it carries no cross-runtime parity claim |
| Kotlin | JVM/Android client substrate | Typed MessagePack cache, one-call typed snapshot document sync into the receiver's own registered type for the schema, built-in schema catalog advertisement, schema/shard catalogs with shared schema-message transport fetch/sync helpers, authority leases, authorized peer lookup, stream catalog, branded `CultMesh` facade, WebSocket transport profile/connection helper, RUDP client/server sugar, connected RUDP CultMesh client helpers, Kotlin interop RUDP serve/dial path, RUDP reconnect policy/controller and reconnect-loop helper, schema-v0 MessagePack over RUDP evidence | C# SoA store (parked at `parked/cultcache-soa`), full C# server daemon, or Unity hot-loop memory layout |
| **C#** | [first call, tested](#c-first-call) | [package](../../packages/cultmath/README.md); snippet not yet | [package](../../src/GameCult.Geometry/README.md); snippet not yet | [first call, tested](#c-cultnet-first-call); [package](../../src/GameCult.Networking) | [quickstart](../../src/GameCult.Mesh/docs/durable-node-quickstart.md); run by `scripts/verify-eve-getting-started.ps1` |
| **TypeScript** | [first call, tested](#typescript-first-call) | not yet | not yet | [first call, tested](#typescript-cultnet-first-call); [package](../../packages/cultnet-ts/README.md) | [package](../../packages/cultmesh-ts/README.md); snippet not yet; [recipe: two-client .io game](io-game.md), tested with two live clients, browser page not yet run |
| **Rust** | [first call, tested](#rust-first-call) | [native core](../../packages/cultmath/native/cultmath-core); snippet not yet | not yet | [first call, tested](#rust-cultnet-first-call); [package](../../packages/cultnet-rs/README.md), [examples](../../packages/cultnet-rs/examples) | [package](../../packages/cultmesh-rs); snippet not yet |
| **Python** | [first call, tested](#python-first-call) | not yet | not yet | [first call, tested](#python-cultnet-first-call); [package](../../packages/cultnet-py/README.md) | [package](../../packages/cultmesh-py/README.md); snippet not yet |
For the whole
picture of what each piece is for, see the
[package chooser in the top-level README](../../README.md#which-package-do-i-want).
## C# first call
Define a typed document, put it in a cache backed by a `.cc` file, read it back through a second
The test imports from `../src/index` where the block reads `@gamecult/cultcache-ts`; a second test in that file fails if the block stops matching the rest. More:
`packages/cultcache-ts/README.md`.
## Rust first call
Same shape with `derive(DatabaseEntry)`: slots are `#[cultcache(key = N)]`, so the `.cc` payload is the
slot-indexed array the C# runtime writes. The source is
[`packages/cultcache-rs/tests/first_call.rs`](../../packages/cultcache-rs/tests/first_call.rs), run by
`cargo test --test first_call` in `packages/cultcache-rs`; a second test in that file fails if the block
below stops matching the file. The Rust runtime has no name lookup, so the read-back is by key. More:
# Recipe: a two-client .io game over CultCache + CultMesh (TypeScript)
A tiny arena. One process is the authority: it holds a single `io.arena_state` document in a CultCache
and serves it over CultMesh. Each client discovers the authority through the Odin catalog, leases the
document, and invokes `join`, `move` and `leave` operations. Two clients see each other join, move and
leave because both lease the same document.
## What ran, what did not
| Piece | Status |
| --- | --- |
| Two clients, real WebSockets, real `@gamecult/cultmesh-browser` client, one arena: join, see each other, move, leave | **tested** (`arena.test.ts`) |
| Edge clamp, and the source below matching the files that ran | **tested** |
| Browser bundle builds and the page is served | **tested** (HTTP fetch only) |
| The page running in a real browser, two windows | **not yet**: no browser on the host that built this recipe |
| Run by CI / `npm run test:ts` | **not yet**: the recipe has its own `package.json` (for `ws`) outside the workspaces; run it by hand as below |
| Authority state on disk (`.cc` file) | **not yet**: the arena cache is in memory |
| Remote play: Odin roots and a signed route | **not yet**: the client uses `local-development` trust, loopback only |
| Dropped players: a closed tab that never sent `leave` stays in the arena | **not yet**: operations do not carry a connection identity to the authority |
| `CultCache` in the browser | **not yet**: the `@gamecult/cultcache-ts` entry imports Node built-ins, so the client keeps the decoded document in a variable and only `./document` is imported |
No `@gamecult/cultmesh-ts` or `@gamecult/cultnet-ts` TypeScript authority server exists. The server below is the CultNet
message handling a provider owes, written in this recipe with the `ws` package.
## Run it
```
npm ci --ignore-scripts # repo root
(cd packages/cultnet-ts && npm run build) # then cultcache-ts, cultmesh-browser, same command
cd docs/start/io-game && npm install
npm test # node --test arena.test.ts, Node 24
node serve.ts # http://127.0.0.1:8080, open it twice
```
Tested on Node v24.14, which runs `.ts` files directly (no type check; `tsc` is not part of this recipe).
## schema.ts
```ts
import { defineDocumentType } from "@gamecult/cultcache-ts/document";
The two Self readings disagree on one case: a record under an undeclared older id of the written schema name, read whole
(an undeclared version bump). Literal "anything else is refused" makes it read-only; `drift-record-write-ownership` makes it
owned. That is `variants:question:written-type-undeclared-drift`, and only the operator can settle it.
## Body facts (probed at b7a68884)
Probes: scratch tests in this worktree, removed. C# filtered runs on Starfire (one job each); TS and Rust on Yggdrasil
(`ygg-verify.sh`, scratch commit `78f604fd`, not pushed to the forge); Python locally. Types A (`tests.wo_a`) and B
(`tests.wo_b`) share one shape; N1/N2 are two versions of `tests.wo_n`, nothing declared.
| Probe | C# | TypeScript | Rust | Python |
|---|---|---|---|---|
| P1 cache {A,B}: B at k, then A at k | **C**: B's record replaced, k now A's id | C, and **both** records live at k | C, both live at k | C, both live at k |
| P2 cache {A} only, store holds B at k, write A at k | R `CultSchemaConflictException` (B is foreign, carried) | R at load: "No schema is registered for persisted schema id" | R at load, same | R at load: "Unknown persisted document type" |
| P3 cacheless `WriteDocument<A>` over B, registry null / {A} / {A,B} | R / R / R | no cacheless write | no cacheless write | no cacheless write |
| P4 cache {A,B}: remove k and upsert A at k in one batch | C (k is A's) | n/a | n/a | n/a |
| P5 cache {A,B}: remove, then add in a second commit | C | n/a | n/a | n/a |
| P6 N2 over an N1 record (undeclared name drift), cache {N2} / cacheless reg {N2} | C restamped / C restamped | not probed | not probed | C (cache) |
What this establishes:
- C#: one record per key. The only path that retypes is the cache arm (P1): `CacheBackingStore.RefuseUndeclared`
(`CultCache.cs:3994-4009`) and the on-disk judge (`OwnsOnDisk`, `:4731-4763`) ask `Registry.Owns(schemaId, catalog)`
(`:841-842`), which answers per schema id for the whole registry: any registered type declaring the id makes every
registered type's write to that key owned. The cacheless arm already refuses the retype (P3) but decides it through
`CatalogEntryLookup`'s registry half (`:596-634`) and the claimant precedence, which is what the ruling removes.
The judge is also per batch, not per key: the cacheless predicate's `declared` is the union over every written entry
(`:4752-4754`), and `UndeclaredOnDisk` memoises one answer per schema id (`:4784-4797`).
- Retype as remove plus add already works in the cache (P4, P5); the cut adds no new primitive.
- Directory store: its own judge (`DirectoryMessagePackBackingStore.cs:287-306`) asks `Registry.Owns` for every key,
upserts included. It must take the same per-key rule.
- TypeScript, Rust, Python: records are keyed by `(type, key)` (Rust `entries: BTreeMap<(String, String), _>`,
`lib.rs:2659`; TS `#entryId(definition.type, key)`, `cult-cache.ts:323`; Python sorts and stores by `(type, key)`).
A write of A at k never meets B's record: both live at k (P1). None carries a foreign record; a store holding an
unregistered type does not load (P2). So no leaf has a "write meets another type's record" path to fix. Bringing the
ruling to the leaves first needs C#'s key identity (one record per key) and C#'s foreign carry, which are parity changes
of the leaves' store model, not a write-ownership rule. Owned by the new follow-up `variants:follow_up:leaf-key-identity-parity`.
- Interop (unproven): a Python store holding A and B at k fails C# decode first on payload encoding (Python's
lambda-defined document wrote JSON payloads), so what C# does with two records at one key was not reached.
## Model and authority table
| Value | Today (b7a68884) | After the cut |
|---|---|---|
| May an upsert of T replace the stored record at k? | cache: `Registry.Owns(id)` (any registered type); cacheless: union of written entries' declarations plus claimant over written entries and the lent registry | **Owner: `CultDocumentRegistry.WrittenOwns(T's entry, stored id, catalog)`**, per key, both arms, both store kinds |
| May a removal (or check) at k proceed? | cache: `Registry.Owns`; cacheless: the written-entries predicate | unchanged in kind: cache `Registry.Owns`; cacheless: owned when some entry the call writes `WrittenOwns` it (no registry) |
| Which type reads a record (claimant precedence) | `Claimant<T>` | unchanged; read machinery only, no write asks it except through `WrittenOwns`' one-entry lookup (option `own-entry-reads-whole`) |
| Lent registry in a cacheless write | encodes, decides copied-record v3 header, adds claimants | encodes and decides the copied-record header (`CopiedHoldIds`, `:4600-4630`); **adds no claimants** |
- `msgpack` is a hard dependency, as `pyproject.toml` declared it. The optional-import fallbacks and the `cultcache-py[msgpack]` extra they named (which never existed) are gone: a missing msgpack is an `ImportError` at import.
- The single-file store's `push`, `push_all` and `delete` take the cross-process lock on `<path>.lock` (flock; LockFileEx over the whole file on Windows) across the read and the replace, waiting up to `lock_timeout` seconds (30 by default, `TimeoutError` after), and exclude the C#, Rust and TypeScript runtimes' writers. The staged file is `<path>.tmp-<pid>-<uuid>`, fsynced before the replace, with a directory fsync after it off Windows. A fixed `<path>.tmp` is no longer used.
### Breaking
- A single-file store written as `cultcache.store.v3` is read and written: a record that holds an element id (an owned list element with an `[CultElementId]` member, carried through every write) marks the store v3, and a v3 store is carried through a rewrite with its ids. A runtime from before this change refuses a v3 store by name.
- A write over a file the reader refuses (`StoreUnreadableError`, from `push`, `push_all` and `delete`) refuses and leaves the file as it is. A flush used to replace such a file. A write set cannot merge with a disk it cannot read, so the file is repaired or removed by its owner before the store writes again.
- Added `StoreUnreadableError` (a `ValueError`) for a store file this runtime cannot read, and `SchemaConflictError` (a `ValueError`) for a
write the catalog cannot describe. A store is exactly three slots; a schema is never recovered from a record's payload.
- The single-file writer derives the catalog from the records it writes: one entry per carried schema id, the registered entry over an
arrived one, written as chosen and never merged. A cache stamps a record with its registered schema's id.
- Every registration refusal is now a `SchemaConflictError`, a `ValueError`, where it was a `CultCacheError`, a `RuntimeError`: registering a document type or a schema name twice no longer raises what `except CultCacheError` or `except RuntimeError` catches.
- Two documents that own one schema id are refused at registration (`SchemaConflictError`). 2.0.1 accepted them. A document may still list another's id as compatible.
- A loaded record resolves by its schema id first: the document that owns the id, else the one document that lists it as compatible; several listers and no owner refuse the record with `SchemaConflictError`. The record's type name resolves it only when its schema id names no registered document. 2.0.1 resolved by the record's type name alone.
self.assertGreaterEqual(time.monotonic()-holder.held_at,HOLD_SECONDS-0.1,"the write finished while the holder still held the lock")# type: ignore[attr-defined]
All notable changes to `cultcache-rs` are recorded here, following
[Keep a Changelog](https://keepachangelog.com/en/1.1.0/). Earlier releases are recorded
only in git history.
## Unreleased
### Breaking
- A single-file store written as `cultcache.store.v3` is read and written: a record that holds an element id (an owned list element with an `[CultElementId]` member, carried through every write) marks the store v3, and a v3 store is carried through a rewrite with its ids. A runtime from before this change refuses a v3 store by name.
- A write over a file the reader refuses (a zero-byte file, a truncated store, a stamp this runtime does not read) refuses with `CultCacheStoreUnreadable` and leaves the file as it is. A whole-store write used to replace such a file. This reverses 0.3.1, where `push_all` replaced a file with no readable stamp (healing it); the stamp rule now has one owner, the reader every write asks under the lock, and it accepts v1 and v3 (0.3.1 refused v3).
### Added
- Lookups, matching the C# `GetByName`, `GetByIndex` and `GetGlobal`. `#[cultcache(name)]` and `#[cultcache(index)]` / `#[cultcache(index = "alias")]` on `DatabaseEntry` fields and `#[cultcache(global)]` on the struct give `CultCache::get_by_name`, `get_by_index` and `get_global`. An empty name and a blank index value are not found; records of one type sharing a name or value find the one with the smallest key. A second record of a global type under another key is refused on every write path and on load with `CultCacheSecondGlobal`. Two `name` members, or two members with one index alias, are refused at compile time. A member's text comes from `CultLookupValue` (`String`, `Option` of one, integers, `bool`, `char`).
- `SchemaConflictError`, the typed refusal of a write the catalog cannot describe: records of different types under one schema id, or an arrived schema and a registered one that share an id and disagree on the schema name. A whole-store write (`push_all`) refuses with it, names the schema id, the names involved and a record key, writes nothing and leaves the file as it is. A caller tells it apart with `error.downcast_ref::<SchemaConflictError>()`.
## [0.3.1]
### Fixed
- `push_all` reads the store's header under the lock and refuses a format this runtime does
not read, with `CultCacheStoreUnreadable``UnsupportedFormat`, leaving the file as it was.
A file with no readable stamp is still replaced, so a whole write heals it.
All notable changes to `@gamecult/cultcache-ts` are recorded here, following
[Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
## [0.14.1]
## [0.15.0]
0.14.1 and 0.14.2 were never tagged or published; this entry carries them with the Unreleased work, and `cultcache-ts-v0.14.0` is the previous release.
### Breaking
- A single-file store written as `cultcache.store.v3` is read and written: a record that holds an element id (an owned list element with an `[CultElementId]` member, carried through every write) marks the store v3, and a v3 store is carried through a rewrite with its ids. A runtime from before this change refuses a v3 store by name.
- A single-file store refuses a zero-byte file as `StoreUnreadableError`; only a path with nothing at it is an empty store. A header this runtime does not read is refused, and the refusal echoes the header only as `cultcache.store.v<digits>`.
- A write over a file the reader refuses (`StoreUnreadableError`, from `push`, `pushAll` and `delete`) refuses and leaves the file as it is. A flush used to replace such a file. A write set cannot merge with a disk it cannot read, so the file is repaired or removed by its owner before the store writes again. A full `pushAll` over a file with no readable stamp does not replace it: the reader refuses it. The stamp rule has one owner, `refuseUnreadFormat`, asked only by the one reader that every write consults under the lock. The reader judges the stamp from the header before it decodes the body, so a newer store is refused by name whether or not its body is whole.
- A dangling symbolic link at the store path is an I/O error with code `ENOENT`, not an empty store.
### Added
- `@gamecult/cultcache-ts/browser`, also what a bundler resolves for the package root under the `browser` export condition: the root surface without `SingleFileMessagePackBackingStore` and `readSingleFileStore`, so it imports no `node:*` built-in and no `koffi` and bundles with esbuild `platform: "browser"`. The Node entry exports exactly what it did.
- `readSingleFileStore(path)`, the one synchronous reader of a single-file store, for consumers that must read a store file rather than decode it themselves. `pullAll`, `push`, `delete` and `pushAll` ask the same reader.
- `StoreUnreadableError` for a store file this runtime cannot read, `SchemaConflictError` for a write the catalog cannot describe, and `schemaIdentityOf` with its result type `CultCacheSchemaIdentity` (`schemaId`, `schemaName`, `schemaVersion`, `compatibleSchemaIds`), the one derivation of a definition's identity.
- The document payload codec, exported from `@gamecult/cultcache-ts/document`: `decodeDocumentPayload`, `encodeDocumentPayload`, and the two named byte escapes `documentPayloadFromBytes` and `documentPayloadBytes`.
- `SingleFileStoreRead`, exported: the result type of `readSingleFileStore` (`absent`, `format`, `envelopes`, `repairedLegacyPayload`).
- `CultCacheInspection.format` is `StoreFormat | "legacy.envelope-array"`: it reports `cultcache.store.v3` for a v3 store.
- A soft `pushAll` writes only where nothing is at the path, leaves a store it can read byte for byte, and refuses what the reader refuses.
- `SingleFileMessagePackBackingStore` takes a cross-process write lock on `<file>.lock`,
interoperable with the C# and Rust runtimes (`flock` on Unix, `LockFileEx` on Windows).
`push` and `delete` read the current file, merge and replace it under the lock, so concurrent
TypeScript and Rust writers no longer lose each other's records across those calls. A full
`pushAll` still replaces the whole store, so records a concurrent writer added are lost to it.
C# writes are not all alike: a conditional commit on a clean store lands onto the file as it
is, but C# writers can still overwrite records a concurrent TypeScript writer added.
- The staged file is fsynced before the replace, and on Unix the directory after it. A
directory sync failure after the replace is reported as replaced but not durable.
- `checkPayloadDepth` is re-exported from `@gamecult/cultcache-ts/document`, so dependents
(`cultnet-ts/content`) bound payload nesting with the same check the cache
applies on decode.
### Changed
- The Windows replace is a POSIX-semantics rename first (it lands while C# and Rust readers
hold the store open), falling back to a write-through `MoveFileExW`, with 25 attempts 20 ms
apart on errors 5, 32 and 33. Writers wait up to 30 s for the lock.
- Opening a legacy store that needs healing rewrites it under the lock, from a read taken
Browsers import the same names from `@gamecult/cultcache-ts/document`, which holds no Node built-ins.
Browsers import the same names from `@gamecult/cultcache-ts/document`, which holds no Node built-ins. The whole cache without the single-file store is `@gamecult/cultcache-ts/browser`, which is also what bundlers resolve for the package root (`browser` condition); bring your own `CacheBackingStore`.
## `.cc` Inspection
@ -358,7 +358,8 @@ That is the TypeScript version of the original CultCache trick:
- each record is `[key, schemaId, storedAt, payload]`
- each catalog entry is `[schemaId, schemaName, schemaVersion, contentHash, canonicalSchemaJson, compatibleSchemaIds, members]`
- it still reads the older TS/Rust-style envelope array and rewrites repaired legacy payloads as v1 snapshots
- it queues local writes so one process does not step on its own shoelaces
- it queues local writes in order, then takes the shared `<file>.lock` (`flock` on Unix, `LockFileEx` on Windows, via koffi, loaded on the first write) for each one, so C#, Rust and TypeScript writers exclude each other; the read a write merges from is taken under that lock
- it fsyncs the staged file and, on Unix, the directory after the replace; on Windows it replaces with a POSIX-semantics rename and falls back to a write-through `MoveFileExW`, retrying briefly on sharing violations
- it persists raw MessagePack payload bytes inside schema-addressed records
It is a control-plane store, not a tiny god. If multiple processes hammer the same file, you still need a coordinator or an external lock instead of prayer.
It is a control-plane store, not a tiny god: every write rewrites the whole file.
`CultCache store format ${describeHeader(decoded[0])} is not one this runtime reads; it reads "${STORE_FORMAT_VERSION}" only.`,
`CultCache store format ${describeHeader(header)} is not one this runtime reads; it reads "${STORE_FORMAT_VERSION}" and "${STORE_FORMAT_ELEMENT_IDS}" only.`,
test("the Windows replace falls back to write-through MoveFileExW where POSIX rename is unsupported, retries 5/32/33, and gives up after 25 attempts",async()=>{
/persisted record "rec:1" references missing schema id "sha256:missing"/,
);
}finally{
awaitrm(directory,{recursive: true,force: true});
}
});
test("opening a store that needs no healing writes nothing: a missing store leaves no lock file, a legacy store with byte payloads keeps its bytes",async()=>{
`decay`/`damp`, intercept/segment helpers (`distance_to_segment` in 2D and 3D, with and without `out closest`, and `segment_closest`), Catmull-Rom, Bezier curves, and
deterministic value-noise primitives
deterministic value-noise primitives, and `fbm(p)` / `fbm(p, octaves, lacunarity, gain)`, the value-only octave sum
of `snoise` (`cultmath_fbm` in HLSL and GLSL)
- Site cells: `power_cells` and `multiplicative_cells` return the two nearest sites of a point as `CultSiteCells`,
`soft_gabriel_gate` gates a pair of sites by the empty-disc margin. Each is a loop over the shader steps
<!-- MIT overall; math.Phacelle.cs and CultPhasor.cs (and the two Phacelle HLSL files) are MPL-2.0; math.Cycles.cs and CultMath.Cycles.hlsl are Apache-2.0. -->
<PackageLicenseExpression>MIT AND MPL-2.0 AND Apache-2.0</PackageLicenseExpression>
<Version>0.4.1</Version>
<Version>0.5.0</Version>
<Description>Clean-room C# math shaped for HLSL-style renderer code.</Description>
json.Append(" \"encoding\": \"each case is 'arguments -> result', every scalar a float32 or 32-bit integer bit pattern in hex, vectors and structs flattened in field order\",\n");