Repository navigation
Scope person-query evidence to decision reads - #1072
Merged
Merged
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
This was referenced Nov 20, 2026
zlareb1
force-pushed
the
fix/scim-person-query-capture
branch
from
November 20, 2026 22:17
7cd6fe5 to
7793ec8
Compare
This was referenced Nov 20, 2026
Merged
zlareb1
force-pushed
the
fix/scim-person-query-capture
branch
5 times, most recently
from
November 21, 2026 02:03
298d854 to
7461d8c
Compare
4 of 5 tasks
Shared client schema reads run with an independent context. Record only decision-tagged statements while retaining unbounded decision reads as failures. Signed-off-by: zlareb1 <lareb.zafar@clickhouse.com>
Signed-off-by: zlareb1 <lareb.zafar@clickhouse.com>
zlareb1
force-pushed
the
fix/scim-person-query-capture
branch
from
November 21, 2026 03:21
7461d8c to
68e8f00
Compare
4 of 5 tasks
zlareb1
added a commit
that referenced
this pull request
Nov 21, 2026
Related: #1072 Related: #1025 ### Changelog category (leave one): - CI Fix or Improvement (changelog entry is not required) ### Changelog entry (one or two sentences a user of Loom can read without the diff; the release pull request collects these into the release notes): Not required. ### Why A failed claim cancellation can leave a task with one cancellation request while it still waits for input. The failure trace needs the waiter's state to distinguish a consumed request from a cancelled waiter awaiting its task's next step. ### Scope and risk Only the claim cancellation test's failure observer changes. It records fixed task and waiter booleans, plus the pending cancellation flag when the runtime exposes it. Task names, arguments, local values, input and exception content stay out of the packet. The existing task, frame and byte limits remain. This adds diagnostic evidence; it does not establish or repair the cancellation failure's cause. The command, signal tests, stdin behavior, lifecycle assertions and exit deadlines are unchanged. ### Compatibility, rollout and rollback - Wire: none. - Operator steps: none. - Rollback: reverting leaves no stored data or generated files. ### Tests Two isolated controls fail with missing observer fields before the change and pass after it. They distinguish a cancelled waiter before task resumption from a consumed cancellation followed by a fresh waiter. The five selected observer and privacy cases pass. Scoped Ruff, format, prose and whitespace checks pass. An AST comparison verifies that every existing test and helper except the observer is unchanged, that all original assertions remain, and that the two new controls match their failing versions. Full hosted CI and a fresh independent review are required after publication. The natural signal race and the claim subprocess tests were not repeated locally. ### AI assistance in the code (leave one; writing this description with AI is fine): - AI-generated: Codex wrote the observer fields and isolated controls. The source diff, regression failures, focused results and preservation proof were checked. - [x] I can explain every line of this diff and how it interacts with the rest of Loom. ### Checklist - [ ] Full format, lint, type and unit gates pass in hosted CI before merge. - [x] No generated files, migrations, plugin files, frontend code, runtime configuration or wire behavior change. - [x] Public text passes the prose check. - [x] The commit is signed off. Signed-off-by: zlareb1 <lareb.zafar@clickhouse.com>
4 of 5 tasks
zlareb1
added a commit
that referenced
this pull request
Nov 21, 2026
Related: #1072 ### Changelog category - CI Fix or Improvement ### Changelog entry Not required. ### Why HTTP startup can consume the lookup deadline before a request reaches the local server. The HTTP cases check bounded native fallback for attempted requests. Controlled child fetches check response handling and abort of a request that entered the pending state. ### Scope and risk Two grep test files separate attempted HTTP fallback from admitted response and abort controls. The existing HTTP cases retain their three modes, mode order, native output and exit checks, 300 ms request deadline and 3000 ms elapsed bound. Those cases do not establish that the server received a request. A test-only preload supplies refusal and malformed JSON response objects before the shim deadline begins. A stalled mode validates the fetch arguments and live signal, enters a pending promise, and requires the shim signal to abort with `TimeoutError`. It records zero JSON calls. A separate 3000 ms fixture watchdog rejects if that abort never arrives; the test rejects watchdog expiry. The abort path clears the watchdog, and the watchdog path removes its abort listener. Every mode retains native output, exit status and the 3000 ms elapsed bound. The response controls retain their invocation, request-argument and JSON parsing assertions. Existing successful annotation cases still exercise real HTTP. The historical macOS failure has no transport-phase trace. This fixes the fixture contract without claiming a measured scheduling cause or a production defect. The two reviewed person-query fixture files from #1072 are already present on main. Their integration capture snapshots the task-local decision tag before awaiting either query method, excluding unrelated background reads. Decision reads with no timeout remain visible to the unchanged finite-budget assertion. Ten controls cover concurrent schema reads, both query methods, unbounded decision reads and the real decision reader with a fake driver. Hosted integration CI recorded a 2.5-second captured limit against the 1.75-second ceiling. The log does not identify the offending SQL. The retained fail-first controls reproduce the capture defect; this composition does not establish that run's unique cause or change the assertion. ### Compatibility, rollout and rollback - Wire, schema, settings and operator steps: none. - Production code, plugin manifests, versions and budget limits are unchanged. - Rollback: revert the two grep test files. No stored data changes. ### Tests The current-main composition keeps the existing prerequisites. Only the two grep test files change after the main merge. No previous CI result or review verdict qualifies this head. Two focused Node cases pass: the attempted HTTP fallback case and the controlled child case covering refusal, malformed JSON and an admitted stalled fetch. The stalled-fetch control first fails against the unsupported fixture, then passes with the abort path. That failing control demonstrates missing deterministic coverage, not a production defect or the historical hosted cause. Node syntax, formatting, prose and committed-range whitespace checks pass. Full hosted CI and a fresh independent whole review remain required. Earlier hosted Go-cost and translation-worker failures remain recorded; this fixture change does not establish their causes. No live provider, benchmark, runtime or CI retry was run. ### AI assistance in the code - AI-generated: Codex investigated the fixture and wrote the response controls. - [x] I can explain every line of this diff and how it interacts with the rest of Loom. ### Checklist - [x] Native output, exit status, timeout coverage and finite-budget assertions are retained. - [x] No production, generated, wire or configuration changes. - [x] Commits are signed off. - [ ] Full required exact-head hosted CI passes. Current-main composition preserves independent human administrator approval for proposals and the stored review-provenance field. Focused local checks on this source are recorded separately; fresh whole-source review and complete exact-head hosted CI are required. Earlier CI failures and their unproven causes remain recorded. No previous review or CI pass qualifies this composition. Signed-off-by: zlareb1 <lareb.zafar@clickhouse.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Related: #1045
Related: #1100
Changelog category
Why
The person-decision integration fixture wraps a shared client and records unrelated background reads. A schema watcher uses an independent context and can omit a default read timeout. Capture only the decision query tag observed at entry, while retaining decision reads with no timeout so the finite-budget assertion can reject them.
Scope and risk
Both query methods retain the original finite-budget assertion. Ten controls cover concurrent schema reads, unbounded decision reads and the real decision reader with a fake driver. The earlier hosted log does not identify its offending SQL; these controls reproduce the capture defect without establishing a unique cause for that run.
The two repository renewal cases use a private function loop whose watchdog clock advances during selector waits. Repository expiry stays separate. Original bodies, the 5 ms expiry, 100 ms watchdog, cancellation cleanup and base-proof assertions remain.
The claim-cancellation test module composes the reviewed diagnostic observer from #1100. On failure it records fixed task and waiter booleans within the existing privacy and byte limits. Two event-driven controls distinguish cancellation awaiting task resumption from cancellation consumed before another pending wait. Original claim assertions, input handling, signal delivery and deadlines remain unchanged. The natural cancellation failure's cause remains unknown; this adds evidence and does not repair cancellation.
Compatibility, rollout and rollback
Only test fixtures and their controls change. Production code, wire fields, settings, schema and operator steps remain unchanged. Reverting restores the previous test fixtures. No stored data changes.
Tests
On this current-main head, 36 directly related SCIM query, replicated-erasure and repository-expiry cases pass. Five pure claim observer and privacy controls also pass; no real claim subprocess or signal-race test ran locally. Scoped Ruff, formatting, public-body prose and committed-range whitespace checks pass.
Earlier fail-first capture and observer receipts remain prior-source evidence. They are not new runs on this head. All three prior concern files retain their exact reviewed bytes and modes; the claim module exactly matches #1100. The complete tree is current main plus these four files. Full hosted CI, including the actual-client integration and claim cancellation cases, and a fresh independent whole review remain required for this exact head.
AI assistance in the code
AI-generated: Codex prepared the fixture corrections and diagnostic controls.
I can explain every line of this diff and how it interacts with the rest of Loom.
Checklist