Commit Graph
20798 Commits
Author SHA1 Message Date
Tianyi Cui 3b00aebf40 Merge pull request #5721 from deepseek-harness/worktree/long-session-paging-observer-20261005
test(bench): exclude paging locator preparation from timing
2026-10-05 15:27:43 +08:00
Tianyi Cui 537164af0c Merge pull request #5714 from deepseek-harness/worktree/model-form-hydration-flake-20261005
fix(models): prevent stale actions during settings refresh
2026-10-05 15:25:51 +08:00
Tianyi Cui c2a5b56fb5 Merge pull request #5711 from deepseek-harness/worktree/wheel-deflate-comparison-flake-20261005
test(infra): compare wheel bytes without deep matcher traversal
2026-10-05 15:17:45 +08:00
Tianyi Cui c7d3a8397b Merge pull request #5715 from deepseek-harness/worktree/trajectory-locale-readiness-20261005
test(trajectory): await locale readiness in bundle fixture
2026-10-05 15:14:46 +08:00
Tianyi Cui b885088946 test(bench): wait for paging controls with isolated clocks 2026-10-05 15:12:09 +08:00
Tianyi Cui bb9dfc663b test(bench): resolve paging controls before timing 2026-10-05 14:25:49 +08:00
Tianyi Cui e994542a31 test(models): use typed provider failure mock 2026-10-05 13:32:53 +08:00
Tianyi Cui c880291a0b fix(models): wait for current snapshot when retrying setup 2026-10-05 13:19:22 +08:00
Tianyi Cui bd20fc56bb fix(models): gate stale provider actions during refresh 2026-10-05 13:09:14 +08:00
Tianyi Cui ef29f117da test(trajectory): await locale readiness in bundle fixture 2026-10-05 12:08:10 +08:00
Tianyi Cui 1fb43c6f78 fix(models): await current snapshot before reopening editors 2026-10-05 11:54:31 +08:00
Tianyi Cui d32ecfe1e4 test(infra): compare wheel bytes without deep matcher traversal 2026-10-05 11:46:52 +08:00
Tianyi Cui f230806884 Merge pull request #5705 from deepseek-harness/worktree/snapshot-deadline-flakes-20261005
test(snapshot): retain goal and turn-start wait diagnostics
2026-10-05 10:55:24 +08:00
Tianyi Cui ec73ee43b4 Merge pull request #5704 from deepseek-harness/worktree/fix-seeded-history-clock-20261005
test(web): pin seeded history to its fixture day
2026-10-05 10:47:16 +08:00
Tianyi Cui 88796a9c31 Merge pull request #5698 from deepseek-harness/fix/credentials-watcher-close-20261005
fix(credentials,hmr,skills): contain watcher errors that arrive after close
2026-10-05 10:46:23 +08:00
Tianyi Cui 910f97059f test(snapshot): retain goal and turn-start wait diagnostics 2026-10-05 10:30:04 +08:00
Tianyi Cui 423aad3534 test(web): pin seeded history to its fixture day 2026-10-05 10:27:49 +08:00
Tianyi Cui a790778e8f refactor(watch): name the close promise and keep one mechanism note
Review follow-up: `const closed` beside the boolean `this.closed` reads as the
flag, so name the promise `closing` at every site; the test-side comment
repeated the production mechanism verbatim, so each test keeps only the
ordering assumption its fake encodes.
2026-10-05 10:13:54 +08:00
Tianyi Cui 14e9bf61cb fix(hmr,skills): contain watcher errors that arrive after close
The EPERM path in the Windows coverage failure also matches the HMR
exact-config watcher: it watches the temp harness home with chokidar's
default 2 s write-settle window, and that spec constructs it on the very
directory its afterEach deletes. Apply the same containment to that
watcher and to the skill-filesystem root watcher, the remaining owners
that enable awaitWriteFinish.
2026-10-05 10:12:31 +08:00
Tianyi Cui 7c450b845d Merge pull request #5699 from deepseek-harness/worktree/inbox-timeout-flake-20261005
test(snapshot): retain inbox wait timeout diagnostics
2026-10-05 10:12:18 +08:00
Tianyi Cui a78f0ee85b Merge pull request #5692 from deepseek-harness/worktree/task-relative-clock-flake-20261005
test(schedule): pin task-page relative time assertions
2026-10-05 10:06:25 +08:00
Tianyi Cui f9719330db Merge pull request #5697 from deepseek-harness/worktree/fix-dev-tools-reload-20261005
test(web): await Session restoration before preset menu checks
2026-10-05 09:55:11 +08:00
Tianyi Cui 8a532bc5a5 test(schedule): remove stale wall-clock drift comment 2026-10-05 09:51:35 +08:00
Tianyi Cui df41b7cd2d fix(credentials-local): contain watcher errors that arrive after close
chokidar's close() drops every listener while a scheduled write-settle poll
can still run. That straggler stats a document the home teardown is
deleting, which Windows reports as EPERM, and an 'error' emission with no
listener is rethrown as an uncaught exception: the whole coverage
partition fails while every test file passes.

Re-attach the provider's error listener across the close so the straggler
stays inside the provider, and make it a no-op after disposal.
2026-10-05 09:38:53 +08:00
Tianyi Cui 22f034bd76 test(snapshot): retain inbox wait timeout diagnostics 2026-10-05 09:38:05 +08:00
Tianyi Cui c2f7722d86 test(web): await Session restoration before preset menu checks 2026-10-05 09:37:53 +08:00
Tianyi Cui f8502bb2d2 Merge pull request #4125 from deepseek-harness/worktree/benchmark-trajectory-observer-20260913
fix(benchmarks): sample Trajectory readiness on animation frames
2026-10-05 09:36:09 +08:00
Tianyi Cui 534acc5952 Merge pull request #3649 from deepseek-harness/fix/packed-install-private-npm-cache
fix(release): isolate packed-install npm cache
2026-10-05 09:23:14 +08:00
Tianyi Cui 19e1be4006 test(schedule): pin task-page relative time assertions 2026-10-05 09:21:47 +08:00
Tianyi Cui 51c5d6cd04 Merge master f74d0741 to resolve remaining Trajectory pairing conflict 2026-10-05 09:20:27 +08:00
Tianyi Cui c1b0e84b95 Merge pull request #5079 from deepseek-harness/worktree/python-live-smoke-ack-20260923
test(python): accept standalone live smoke acknowledgements
2026-10-05 09:18:47 +08:00
Tianyi Cui 21507982e6 Merge master checkpoint a2ac28a6 into Trajectory observer fix 2026-10-05 09:10:46 +08:00
Tianyi Cui d2d8c47b4a Merge pull request #5459 from deepseek-harness/worktree/installer-root-validation
fix(desktop): reject canonical drive roots in installer
2026-10-05 09:10:39 +08:00
Tianyi Cui 454c6bfa83 Merge pull request #5620 from deepseek-harness/test/dynamic-tool-cache-e2e-flake-2d600396
test(sdk): keep the dynamic tool cache e2e tolerant of extra model calls
2026-10-05 09:10:17 +08:00
Tianyi Cui d6b2430df1 Merge master checkpoint a2ac28a6 into packed-install cache fix 2026-10-05 09:09:23 +08:00
Tianyi Cui 3ca9334868 Merge observed master checkpoint to resolve Python smoke pairing conflict 2026-10-05 09:05:54 +08:00
Tianyi Cui ef9944a01f Merge pull request #5625 from deepseek-harness/worktree/native-handler-fixture-01a0f9ff
test(native-command): publish Windows handler marker atomically
2026-10-05 09:04:37 +08:00
Tianyi Cui 7f83658b4a fix(desktop): reject canonical drive roots in installer 2026-10-05 08:53:27 +08:00
Tianyi Cui 22e627169e Merge branch 'master' into test/dynamic-tool-cache-e2e-flake-2d600396 2026-10-05 08:50:31 +08:00
Tianyi Cui 0f51ba0737 Merge pull request #5629 from deepseek-harness/test/llm-retry-transport-recovery-timeout-c2c45a16
fix(timeout): end an idle demand the transport abandons
2026-10-05 08:45:18 +08:00
Tianyi Cui 6f88bed362 Merge pull request #5689 from deepseek-harness/worktree/pi-ai-1.0.2-20261004
build(llm-pi-ai): upgrade @earendil-works/pi-ai to 1.0.2
2026-10-05 08:07:26 +08:00
Tianyi Cui 21fe89affe build(llm-pi-ai): upgrade @earendil-works/pi-ai to 1.0.2
pi-ai 1.0.1 and 1.0.2 bring model catalog updates and adapter fixes:
Anthropic native tool changes define later tools inline, Bedrock binds
replayed thinking blocks, OpenAI-compatible requests can pick sampling
parameters by thinking level, and "model is at capacity" is retried.
The runtime dependencies change only in @anthropic-ai/sdk (0.124.0 to
0.129.0) and @earendil-works/pi-telemetry (^1.0.2).

1.0.2 still re-parses the whole accumulated tool-argument JSON on every
stream delta (earendil-works/pi#9265), so the patch is regenerated
against the 1.0.2 sources: it removes the same six per-delta parses as
the 1.0.0 patch, and only the bedrock-converse-stream.js hunk moves, by a
line.
2026-10-05 06:40:08 +08:00
Tianyi Cui 3d0e274859 Merge pull request #5659 from deepseek-harness/worktree/pi-compat-notes-20261003
docs(agent-notes): propose Pi extensions bridge and Pi plugin support levels
2026-10-04 22:35:30 +08:00
Tianyi Cui b3f8bec8e8 docs(agent-notes): address review of the Pi compatibility notes
State the Pi-to-DSH direction of the built-in tool translation, how setActiveTools treats Agent-scoped extension tools, that module-scope extension state is shared by concurrent Sessions, which format edges retain plugin records, and that SessionStartSource already declares clear and compact. Date the npm package count and narrow the plugin note's opening sentence.
2026-10-04 05:08:15 +08:00
Tianyi Cui 61cbe84624 docs(agent-notes): link the issue tracking each core gap
Add the GitHub issue number to each row of the Pi extensions bridge note's core-gap table, and state which rows follow from DSH rules.
2026-10-04 04:52:14 +08:00
Tianyi Cui 6500cacbb2 docs(agent-notes): propose Pi extensions bridge and Pi plugin support levels
Record two proposals from the Pi 1.0.0 compatibility investigation.

The Pi extensions bridge note states the design of an experimental bridge
that loads unmodified Pi extension sources on existing DSH extension
points, and ranks the DSH core mechanisms that are narrower than Pi's
extension API.

The Pi plugins note defers a compatibility layer for Pi's Chord-facet
plugins, whose API upstream declares unstable, and defines six levels of
support with the upstream trigger and the DSH work for each, plus the
runtime differences the plugin model exposes.
2026-10-04 04:52:14 +08:00
Tianyi Cui e1878e41ca fix(llm-pi-ai): withhold ChatGPT sign-in and serve one catalog
pi-ai 1.0.0 adds Sign in with ChatGPT to the `openai` provider, and that
login refuses to start unless the app supplies a stable installation UUID
through LoginOptions.getDeviceId. The harness keeps no installation ID
meant for a third-party account sign-in, so the route offers its API-key
login only; a Loader-composed test pins the offered methods.

catalogProvider(), catalogModels(), and catalogProviderIds() now read one
index that leaves out providers without a chat model, so a hand-declared
route under such a key inherits nothing from pi-ai. The browser-worker
stub gains the builtinModels refusal llm-pi-ai imports and drops the
getBuiltinProviders reader nothing imports. The tool-argument streaming
spec also covers the patched Anthropic Messages path.
2026-10-04 04:51:19 +08:00
Tianyi Cui abfa6302b7 fix(llm-pi-ai): keep catalog providers whose catalog lists no models
The chat-model filter left a provider out whenever pi-ai listed no chat
model for it. The browser-worker preview replaces pi-ai with a stub that
lists no models at all, so every provider was dropped and the empty
directory registration failed the llm-pi-ai mount. Only a provider whose
catalog lists models but no chat model is left out now; the stub gains
the `getAllBuiltinModels` reader the filter calls.
2026-10-04 02:09:17 +08:00
Tianyi Cui d5161807c8 fix(llm-pi-ai): offer only catalog providers that ship a chat model
pi-ai 1.0.0 adds classifier and image models to the built-in catalog and
ships `typesafe` with classifier models only. The adapter dispatches chat
requests only, so a catalog provider without a chat model gets no entry in
the configurable-provider directory and no sign-in flow.
2026-10-04 02:09:17 +08:00
Tianyi Cui f1f18bce98 build(llm-pi-ai): upgrade @earendil-works/pi-ai to 1.0.0
Move the pi-ai provider adapter from 0.87.1 to 1.0.0 and re-create the
local tool-argument patch for the new dist files. pi-ai 1.0.0 still
re-parses the whole accumulated argument JSON on every stream delta in
the Anthropic, Bedrock, Mistral, OpenAI Completions, OpenAI Responses,
and pi-messages adapters (upstream earendil-works/pi#9265), so the patch
removes the same six per-delta parses; arguments are still parsed once at
toolcall_end.

The release-age exemption moves to the exact 1.0.0 pi-ai and
pi-telemetry versions, and the generated third-party notices and README
links name the new patch file.
2026-10-04 02:09:17 +08:00