Commit graph

2,728 commits

Author SHA1 Message Date
Meta Vens
2db8d72e35 Caching docs: Cache is published once, after the pull, so a call makes at most two passes
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 18:05:02 +02:00
Meta Vens
1ccfe615bc Caching tests: churn readers run under a role that does not wait in the reader
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 18:03:09 +02:00
Meta Vens
f574a5a25e Caching tests: the attach lends the codec, Cache is published after the pull, Gate's passes are bounded
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 18:01:30 +02:00
Meta Vens
1468a83a15 Caching: the attach lends the codec with the registry and publishes Cache only after the pull
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>
2026-10-23 18:01:29 +02:00
Meta Vens
a616431ec5 Caching: the gate's stale-guess cap is 1,000, so a validation that runs the body again and again throws before the suite times out
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 17:05:12 +02:00
Meta Vens
ece44a84be Caching docs: the store-composition contract says what the cache gate order enforces today and names the cross-cache read follow-up; CHANGELOG: a failed attach takes its registry back
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 16:34:06 +02:00
Meta Vens
ba6589650f Caching: a failed attach takes its registry back (one UseRegistry writer), and a gate retry that never settles throws instead of looping
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 16:33:01 +02:00
Meta Vens
a2b8d7d28d Caching tests: a failed attach leaves no registry, the Soul P4 load pins the refusal and body in one hold, every cache gate entry and an own-store call keep the order rule (red: failed attach)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 16:33:01 +02:00
Meta Vens
751d9f34e0 Caching tests: a reader queued before an attach runs under the cache's gate after it; an attach pull may read its own cache; user code may call its own store
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>
2026-10-23 14:15:42 +02:00
Meta Vens
d676e4a334 Caching tests: the gate tests dispose their caches after the joins, so a regression fails a bounded join instead of hanging in Dispose
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 14:02:40 +02:00
Meta Vens
3c5f1fe0c2 Caching docs: the lock order is cache gate, store gate, file lock; the gate owner re-checks after locking (contracts, CHANGELOG)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 14:01:43 +02:00
Meta Vens
1eb517296b Caching tests: the PullAll-after-failed-attach test queues on the second attach; a cacheless PullAll without a registry is not a refusal
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:57:13 +02:00
Meta Vens
577526bb28 Caching: a thread holding a store's gate enters no cache gate it does not hold and no second store
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>
2026-10-23 13:54:53 +02:00
Meta Vens
fba7ac1e8a Caching tests: the write-through-a-getter test uses AddAsync (CultCache has no Upsert)
The previous commit's tests do not build.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:45:32 +02:00
Meta Vens
45529db85d Caching tests: user code under a store's gate cannot enter a cache or a second store, and a queued reader runs under the right gate (red)
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>
2026-10-23 13:41:31 +02:00
Meta Vens
9c8973a73e Caching: Cacheless is private protected, so SingleFileBackingStore's WriteRecord and WriteDocument reach it
The previous commit does not build: Cacheless was private to CacheBackingStore.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:38:51 +02:00
Meta Vens
b348978d31 Caching: one owner of store gate selection; delete ForTheCall, RefuseWhenAttached and Held's unchecked choice
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>
2026-10-23 13:38:26 +02:00
Meta Vens
36b22931a3 Gate: every csharp stage runs and the gate names the stages that failed
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:29:53 +02:00
Meta Vens
89d170344a ref-list-drift: the store the 1.0.60 writer makes for a list of record references, written on Yggdrasil at cultlib-unity-v1.0.60
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:28:33 +02:00
Meta Vens
c9da27b28c Gate: the ref-list summary names records by schema name; a flush carries an untouched record's schema id forward here and restamps it at 2.0.1
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:27:18 +02:00
Meta Vens
8efffc1038 Gate: the 1.0.60 writer keys its records alpha, beta and list, so the reference list resolves
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:23:57 +02:00
Meta Vens
7b5f8ca7c3 Gate: the 1.0.60 writer names its types as RefListDriftTests does, and the directory vector keeps the lock its writer leaves
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:22:51 +02:00
Meta Vens
79ed75bedc Gate probe: the 2.0.1 ref-list probe reuses 2.0.1's own RefListDriftTests document types
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:21:18 +02:00
Meta Vens
6dd3ebd0d4 C2a gate vectors: the v3 single-file and v5 directory stores C2aGateWriter writes, written on Yggdrasil at de78734f
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:20:00 +02:00
Meta Vens
de78734f4d C2a format gate: the gate harness, a C# gate probe, leaf probes on the C2a-written vector, and the 1.0.60 ref-list probes
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>
2026-10-23 13:18:57 +02:00
Meta Vens
e06d5a2e8c CachelessSingleFileTests: the second cache's record is keyed by its name's hash, so assert the file's three records, not the key "c"
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 12:21:38 +02:00
Meta Vens
e38925e249 Caching: the cacheless call reaches the store's gate through WhileDetached (the base class owns the gate; 694e6fe6 did not build)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 12:18:12 +02:00
Meta Vens
8b780a27bc cultcache-py CHANGELOG: registration refusals are SchemaConflictError, one owner per schema id, records resolve by schema id first
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 12:17:32 +02:00
Meta Vens
694e6fe632 Caching: a cacheless call takes the store's own gate and refuses under it; the attach holds that gate across its whole body
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>
2026-10-23 12:16:44 +02:00
Meta Vens
748b2d8082 Caching tests: a call arriving during an attach waits for it, success or failure; a refused attach leaves the store cacheless (red)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 12:12:27 +02:00
Meta Vens
5ea97b44a7 CachelessSingleFileTests: WriteDocument plans ids by the declared type (S9)
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-23 04:14:02 +02:00
Meta Vens
792f14bbef CHANGELOGs: cultmesh-ts C2a breaks, cultcache-ts exported types, Unity ForeignRecords and AsSchemaAlias interface removal
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-23 04:11:54 +02:00
Meta Vens
38dfa60804 cultcache-ts: the header peek is the one judge of a stamp; isStoreSnapshot is a shape test; byte-offset views, array16/32 and str16/32 leads and a spaced stamp pinned
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-23 04:11:53 +02:00
Meta Vens
13706c795f Cacheless calls and attach share the store's gate: an attach waits for a call, a call that finds the cache is refused
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-23 04:07:10 +02:00
Meta Vens
8619883b96 cultcache-ts tests: an empty array before a newer stamp is not a stamp
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>
2026-10-23 02:01:05 +02:00
Meta Vens
a4e70ac4fb Cacheless calls detach their registry when they end; the declared type decides the header; both pinned
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>
2026-10-23 02:00:16 +02:00
Meta Vens
3704ce097c CachelessSingleFileTests: the gated store reads the attached registry without requiring one (7d25d92b did not pass: the gate test threw)
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-23 01:49:31 +02:00
Meta Vens
7d25d92bd7 CachelessSingleFileTests: a call keeps its registry while another call waits at the gate
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-23 01:46:09 +02:00
Meta Vens
a056918be6 CHANGELOGs list the measured public surface against 2.0.1 (apicompat strict): members on their owning types, Id members in Mesh, Networking and Geometry, Rust SchemaConflictError, cultnet-rs register
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>
2026-10-23 01:44:57 +02:00
Meta Vens
d1dfd8f027 cultcache-ts: the reader judges the stamp from the header before it decodes the body; the header-form pins return
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-23 01:38:16 +02:00
Meta Vens
19f3a5edbb Cacheless single-file API: WriteDocument encodes the declared type, the registry is the call's argument, DecodeDocument moves to CultCodec, ReadStoreSnapshot's gate is pinned
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>
2026-10-23 01:38:16 +02:00
Meta Vens
f54472e5fe Merge origin/main (cultcache-rs 0.3.1, the whole-store writer fix) into the C2a landing branch
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>
2026-10-23 00:44:59 +02:00
Meta Vens
baee4b6ba1 CachelessSingleFileTests: the attached-store refusal test uses a registered document type; storedAt round-trip format
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-23 00:32:22 +02:00
Meta Vens
2f17721eba CachelessSingleFileTests: WriteDocument and DecodeDocument pin storedAt, random versus deterministic ids, null arguments and the attached refusal before the registry
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-23 00:29:03 +02:00
Meta Vens
02cf04b5c9 Merge hands/whole-store-writers-read-header: Rust push_all and TS pushAll read the header under the lock and refuse a newer store format (cultcache-rs 0.3.1; Soul s2 holds) cultcache-rs-v0.3.1
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 00:17:07 +02:00
Meta Vens
bf01212f9a Docs: cacheless single-file access names WriteDocument and DecodeDocument; CultElementIds is not public
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-23 00:15:28 +02:00
Meta Vens
729754278a Caching: SingleFileBackingStore.WriteDocument and DecodeDocument own element-id planning; CultElementIds is internal again
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>
2026-10-23 00:15:27 +02:00
Meta Vens
6463f34203 CachelessSingleFileTests: null arguments, the lock wait, and the registry deciding the header for copied records
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>
2026-10-23 00:07:30 +02:00
Meta Vens
bd40c05dff cultnet-ts: pin the text() sanitiser around a chunk's hash and id
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>
2026-10-22 23:58:08 +02:00
Meta Vens
74bcf075a4 PreCut2 tests: the v6 suffix refusal asserts the counted suffix, not the absence of the readable stamps it also names
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-22 23:57:51 +02:00