Fail loudly when evaluation answers and questions diverge in count - #1674
Open
oliverames wants to merge 1 commit into
Open
Fail loudly when evaluation answers and questions diverge in count#1674oliverames wants to merge 1 commit into
oliverames wants to merge 1 commit into
Conversation
generate_eval_report pairs each QA pair with its result through zip(). If the two lists ever differ in length, zip silently truncates and the report quietly drops the unmatched rows, which reads as a clean run with fewer questions. strict=True turns that case into a visible error instead. zip(strict=True) needs Python 3.10; 3.9 has been end of life since October 2025. Happy to switch to an explicit length check if you would rather keep 3.9 compatibility.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
generate_eval_reportin the mcp-builder skill pairs each QA pair with its result throughzip(). The two lists should always match one to one, but if they ever drift apart,zipsilently truncates and the report drops the unmatched rows. A broken run then reads as a clean run that happened to have fewer questions, which is the worst way for this script to fail because the person evaluating a description trusts the row count.This passes
strict=Trueso a length mismatch raises instead. It is the same pairing every other consumer of these two lists assumes.One trade-off worth naming:
zip(strict=True)needs Python 3.10, and 3.9 has been end of life since October 2025. If you would rather keep 3.9 compatibility, an explicit length check before the loop works just as well, and I am glad to switch it.I left
webapp-testing/scripts/with_server.pyalone even though it has a similar barezip, because it already validateslen(args.servers) != len(args.ports)and exits before reaching it.