Repository navigation
fix(evals): wait for active native execution before grading - #209
Merged
Merged
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
OpenCode can still be executing while its next model response has not appeared in exported messages. The evaluator could mistake that gap for quiet and grade unfinished work. It now includes the native active-session state for the root and owned descendants in its existing busy predicate, while retaining the same deadline and cancellation paths.
The bug reproduces with a scripted loopback provider. The retained paid run did not capture active state at its stop, so its exact causal connection remains unproven. Its original failed grade and receipts remain unchanged.
Proof: canonical push preflight passed 5,761 tests, with 27 skips and zero failures. Pinned Bun 1.4.0 evaluator checks pass 686 tests with four skips. The same native OpenCode 2.1.4 smoke assertions fail on base while the session is active and pass on head (128 assertions). The packaged plugin remains byte-identical at SHA-256 7883cfbf… .
Reviewed with sol-loop: 1 round with gpt-6.1-sol on high. Sol verdict: approved.
Created with gpt-6.1-sol in T3 Code through the Codex harness.