You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
test(apple): distinguish captured focus words from synthetic types - #4087
Mark the editing and not-editing bridge inputs as rule cases. Their trait words were captured from a SwiftUI TextEditor, but automation type 57 was synthetic in the original focus test. Per-row provenance now records that split explicitly.
One JSON fixture changed. Numeric inputs, expectations, production code and test counts are unchanged. Follow-up to the provenance review of #4073; part of #4045. No live XCTest parity claim.
Validation
Head: 904c91b3aec56d9869ae535a305dedf8815a81f4.
Frozen install/build, full format and all 103 tests across the four producer suites passed. Exact affected validation failed two daemon-entrypoint tests: idle registration remains and startup sweep times out (16,445 passed, two failed, one skipped; run-E3Ovk1, exactCommit: true). Draft pending #4086; unchanged-base full unit passed, but neither failure reproduced, so cause remains unproven. Selected native/replay lanes are CI-owned or parked; local fixture passes do not replace them.
Historical Apple Smoke Tests at 2fbcbb2a2a failed smoke:automation-input while closing the fixture sheet: selector_not_found for automation-close-sheet. That live flow does not consume this fixture; its four readers are unit tests. No infrastructure attribution or smoke-repair claim.
The change looks good at 904c91b. It only changes a test fixture JSON, with no production code, and all 17 checks are green. There are no conflicts. I did not run the test suite; I judged the change by reading the loader, and I cannot check the #3163 capture source from the repo. Not blocking, take or leave: the new per-row provenance key is not in the fixture type in tree.test.ts, so nothing asserts it and it is free text. You could add provenance?: string to that type and assert it is non-empty for rule rows that carry captured words.
On the optional provenance-validation note: I kept the explanatory text outside the decoder's test type. Adding an optional field to a JSON.parse assertion would not validate it at runtime. A non-empty-string assertion would check prose presence, not whether a trait word was captured, and would not detect the original incorrect provenance claim. The executable distinction remains basis: rule; the note explains the mixed captured/synthetic input for human review. No new metadata-only guard or capture-truth claim is added. This PR has since merged; its historical failed local receipt remains attributed to its own head and is not replaced by #4090's later pass.
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
ready-for-humanValid work that needs human implementation, judgment, or maintainer merge
1 participant
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.
Summary
Mark the editing and not-editing bridge inputs as rule cases. Their trait words were captured from a SwiftUI TextEditor, but automation type 57 was synthetic in the original focus test. Per-row provenance now records that split explicitly.
One JSON fixture changed. Numeric inputs, expectations, production code and test counts are unchanged. Follow-up to the provenance review of #4073; part of #4045. No live XCTest parity claim.
Validation
Head:
904c91b3aec56d9869ae535a305dedf8815a81f4.Frozen install/build, full format and all 103 tests across the four producer suites passed. Exact affected validation failed two daemon-entrypoint tests: idle registration remains and startup sweep times out (16,445 passed, two failed, one skipped;
run-E3Ovk1,exactCommit: true). Draft pending #4086; unchanged-base full unit passed, but neither failure reproduced, so cause remains unproven. Selected native/replay lanes are CI-owned or parked; local fixture passes do not replace them.Historical Apple Smoke Tests at
2fbcbb2a2afailedsmoke:automation-inputwhile closing the fixture sheet:selector_not_foundforautomation-close-sheet. That live flow does not consume this fixture; its four readers are unit tests. No infrastructure attribution or smoke-repair claim.