Repository navigation
test(e2e): a native answer's audio may arrive before its closing transcript delta - #1314
Merged
Merged
Conversation
…script delta
A combined-build S106 run failed "timed out after 60000 ms waiting for
haul_e5 input_final, then assistant_audio_start and assistant_audio_end":
the answer ("Kestrel.", 400 ms) started on the media track at 11910 ms, and
its first transcript delta, which closes the input final, arrived on the data
channel at 11926 ms. The wait required the audio start after the final's
entry, so the answer that had already played was never matched.
native_question now takes the first assistant audio after the question's
speech ended, as SpokenTurn already does: the same cross-stream ordering the
word oracle stopped depending on.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why
A combined-build S106 run failed with "timed out after 60000 ms waiting for haul_e5 input_final, then assistant_audio_start and assistant_audio_end". The answer ("Kestrel.", 400 ms) started on the media track at 11910 ms. Its first transcript delta, which closes the question's input final, arrived on the data channel at 11926 ms.
native_questionrequired the assistant audio start to come after the final's timeline entry, so it never matched the answer that had already played. This is the same cross-stream ordering issue that #1310 took out of the row oracle; this fix was pushed to #1310's branch after that PR merged, so it lands here instead.What
native_questionnow waits for the question's input final and the first assistant audio after the question's speech ended, asSpokenTurn::from_timelinealready does. Every scenario that asks a native question uses this helper.Evidence
generate-bazel-rust-builds.mjs --check,machine-check-drift,protocol-check-drift, the poster--check, andcargo fmt --check.🤖 Generated with Claude Code