Commit graph Vili/scripts/test-vili-daemon.mjs
Author SHA1 Message Date
Meta Vens
a9413d4444 Test: Soul s1 findings as behavioural tests, red at the h1 head (job ids, seed, link swap) or pinning gates (lease, LD_LIBRARY_PATH, record links, header time, key loss)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 23:04:36 +02:00
Meta Vens
385a7e5733 Test: the signal comes after the export's first line is logged
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 21:32:26 +02:00
Meta Vens
3d80c7f0b5 Test: wait for the version probe's answer, not its start, before asking for an export
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 21:25:51 +02:00
Meta Vens
1328b5514c Test: a planted directory that is not a job id's serves nothing
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 21:22:48 +02:00
Meta Vens
a65cc60113 Export: the redundant guard after Blender is gone; tests for a lying exit code, a malformed result with an FBX, output after SIGTERM, and a hung request
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 21:17:07 +02:00
Meta Vens
f99d95b4c6 Test: the hung export's grandchild is the one that is new since the orphan's
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 21:08:45 +02:00
Meta Vens
c109f436b8 Tests: malformed result, missing FBX, a grandchild left behind, clip order independent of the worker's listing
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 21:00:22 +02:00
Meta Vens
d4375d458f Tests: say why a job ended before its export; minted seeds differ
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 20:54:48 +02:00
Meta Vens
f38f420df1 The version probe writes nothing to the state root; existing tests learn the new backend and routes
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 20:33:13 +02:00
Meta Vens
9c50b64226 Tests first: the FBX export's behavioural tests and the stub Blender they run against
Red on the daemon at c5cf445: fbx jobs, the refusal ladder, export containment, kill on lease loss and SIGTERM,
the minted seed, motion_job.v1, the artifact read route and rest-fbx. test_bvh_to_fbx.py runs the converter in
real Blender (VILI_BLENDER). This commit does not pass its own tests.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 20:26:54 +02:00
Meta Vens
c094bb0072 Test first: the entry hands cultnet-ts the candidate bind as the Expected route's endpoint (first Idunn run: bind differs)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 14:25:14 +02:00
Meta Vens
6950434f9a Test first: a failed startup names its cause and repeats no value
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 14:17:57 +02:00
Meta Vens
970218c968 Tests: the constraints.json write is gated; a not-yet-existing root named through a symlink is one root (s3 residue)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 13:36:04 +02:00
Meta Vens
1b0d1a2d56 Test: the loss is made after the service-document rewrite has finished, so only the daemon's own ask can notice it
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 05:26:27 +02:00
Meta Vens
74ddacf53b Tests: a worker that ignores SIGTERM is dead within a second of a loss; its output while the daemon stops is not copied
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 05:19:08 +02:00
Meta Vens
1c412908e7 Tests first: loss noticed while a worker is awaited, SIGTERM gates writes and spawns, one root identity, leaderless group, every output line gated
Red at 658de07 for the lease-loss bound, SIGTERM and spelling tests; the leaderless-group and later-line tests pin behaviour that already holds, for mutants.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 05:15:43 +02:00
Meta Vens
658de07927 Test: the daemon's copy of a worker's stdout is gated like every other write
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 03:57:19 +02:00
Meta Vens
b4a4e770cf Test: the request.json write is gated, with the lease lost the moment the worker is healthy
The stub marks its first healthy answer; the stdout line it prints before ready
had been taking the lease through the log-sink gate first.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 03:57:06 +02:00
Meta Vens
f3646e0dfd Tests: the lease lost between any two writes of one request, and during the first persist
The fake Idunn signer takes the lease the moment a named path exists under the
state root. Each gate in generate and startWorker now has a case that only that
gate stops, and a lease lost while the state root opens leaves the daemon up.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 03:33:28 +02:00
Meta Vens
f69e24eeff Test: the spawn is gated at the spawn, not by the request's earlier checks
The fake Idunn signer can lose the lease once a job's meta.json exists, which
pins the guards in startWorker (no logs or embeddings directory, no worker).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 03:06:18 +02:00
Meta Vens
17e84c65da Tests first: crashed-lock race, in-flight fence, SIGTERM stop, foreign reclaim, port range
Red at 4fd3f60 for stale-lock-takeover-race, fence-admits-in-flight-request and
bind-port-65536-accepted; green guards for reclaim-root-unpinned and the SIGTERM
property. The daemon's stop() no longer leaves a daemon holding the run open.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 03:01:17 +02:00
Meta Vens
4fd3f60c88 Test: a second daemon that fails to refuse is killed, not left holding the test run open
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 01:33:22 +02:00
Meta Vens
235321ed86 Entry: derive the daemon's environment in one exported function, and test its argument contract
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 00:45:01 +02:00
Meta Vens
956fdf8ff8 Daemon serves through a runtime port: Idunn write lease or hand-run socket lock
serve(runtime) listens first, writes the state root only once the runtime
holds it, answers Idunn's route challenge ahead of the cross-site refusal,
and fences on lease loss. --health is deleted; the pid file and /proc
reclaim move behind the hand-run runtime. Adds scripts/idunn-entry.mjs,
the kimodo and SOMA-X gitlinks, encode_texts for the child, and tests.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 00:23:07 +02:00
Meta Vens
70f60b9b8a Daemon tests: the pid file's life, strays that ignore SIGTERM, a worker with no encoder child, a live worker left alone
Closes the survivors StrykerJS reported over the new supervisor code (61 mutants, 41 killed, 20 survived): the pid
file is removed when its worker exits and not when it spawned nothing, the restart kill is SIGKILL (a stubborn stray
and a SIGTERM-proof encoder child), the group match is the pid file's, not any marked process. killGroup loses its
unused return value.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 17:54:01 +02:00
Meta Vens
2bf4a1dba7 Daemon: own the worker's whole tree, across worker exits and daemon crashes
The worker's process group is killed when the worker exits for any reason, so no encoder child survives it.
A pid file in the state root names the group; the next daemon start kills a group left by a crashed
predecessor if one of its members carries this state root's embeddings directory, before serving. The
retired operator route's 404 test returns (built from parts, so the grep stays empty) and the test script
gets a per-test timeout.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 17:20:01 +02:00
Meta Vens
e673ec1d4a Test: the retired provider-advertisement route answers 404
Keeps the Eve path out of scripts/ so the negative grep over scripts/ is empty.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 15:28:55 +02:00
Meta Vens
b3ef8f3090 Test: drop the lastError check that no mutant could fail
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 15:22:45 +02:00
Meta Vens
c3b048a2ab Daemon: drop the unread worker.model; tests for grace, host default, relative models root, early death
Mutation triage (StrykerJS, 81 survivors on the first full run) showed worker.model was
written and never read. The new tests pin the survivors that were behaviour: the loopback
default, --host, stray flags, the models-root resolution, the SIGTERM grace, a worker that
dies while the daemon waits for its health, the POST route, and lastError after a
requested stop.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 15:07:15 +02:00
Meta Vens
3843a772f8 Tests: worker failure paths, env twins, group kill, job record, store refresh
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 13:16:50 +02:00
Meta Vens
ef0edc4073 Daemon: supervise the worker child, drop WSL, Docker, Eve and CORS
vili-daemon.mjs spawns the Python worker itself (no shell, port 0 from the
ready line, nice 10, one child at a time, respawn after exit, group kill on
SIGTERM) with a closed environment. WSL/Docker, the Eve surfaces and deck
WebSocket, the provider advertisement, /smoke, CORS, JSON sidecars and the
pid file are deleted. POSTs with an Origin or a non-JSON body are refused
403 cross-site. cultcache-ts comes from the vendor/CultLib submodule at
b549d731; ws is dropped.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-23 12:14:10 +02:00