Skip to content

Preserve native CI failure evidence and verify archived diagnostics - #4683

Merged
realmarcin merged 1 commit into
mainfrom
docs/native-ci-retained-evidence
Oct 8, 2026
Merged

realmarcin merged 1 commit into
mainfrom
docs/native-ci-retained-evidence

Conversation

@realmarcin

Copy link
Copy Markdown
Collaborator

Native CI previously retained too little evidence to distinguish an early observation failure from a deadline failure. This package preserves the original synthetic case archives, matching JUnit archives, GitHub metadata and selected source bytes from PR #4659's run 37759818157, together with a standalone archive verifier and regression controls.

The verifier now separates malformed application-result diagnostics from successful byte retention, bounds local metadata reads, and rejects overflowing JSON numeric exponents. Seven targeted controls passed. Saved byte, source, inventory and JUnit comparisons identify two distinct failures: the neutral case stops on a stream-change refusal at 253.860 seconds; the omission case reaches the existing deadline. The package does not execute native code or claim an established cause, performance improvement or acceptance.

Only evidence notes and standalone tools are added. #4680, #4682, #4537 and #4354 remain open. No runtime guards, acceptance limits or provider behavior change.

Closes #4679.

@realmarcin
realmarcin merged commit bcc0fc9 into main Oct 8, 2026
14 checks passed
@realmarcin
realmarcin deleted the docs/native-ci-retained-evidence branch October 8, 2026 10:58
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Separate native retained-byte verification from partial result diagnostics

1 participant