Repository navigation
drama3: release output ownership on write failures - #14
Draft
cwschroeder wants to merge 1 commit into
Draft
cwschroeder wants to merge 1 commit into
cwschroeder wants to merge 1 commit into
Conversation
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Validation of a two-line cleanup patch on upstream PR ashhart#939 at
3515c27f, which contains main46a08c18.The new fallible WAV-header write can return from
AudioOut.openwhile its FILE is still open. A failedout.close()after a successfulspeakcan also leave the returned host frames without cleanup. Register error cleanup immediately after each acquisition.CPU tests of the actual source blocks reproduce both leaks before the fix and pass all five stdio/ownership cases afterward in Debug and Fast. The three existing write/flush/valid-output controls remain green. Six warmup cases pass in both modes. The actual CLI admission prefix also passes six cases, including codec-only encode without a model; the unchanged original fails that case.
Whole-tree formatting, whitespace and lean comparison pass locally (517 versus 517, zero new). Astra reviewed both cleanup lines without a remaining issue. The full workflow runs here on isolated CI; it has not been rerun locally on this patch.
This draft is for validation and supplying the small patch to the upstream author. It adds no kernel or arithmetic changes and supplies no model, audio-quality or performance qualification.