Repository navigation
fix(agent): ignore tool argument key order - #239
Merged
helloml0326 merged 1 commit intoNov 16, 2026
Merged
Conversation
helloml0326
approved these changes
Nov 16, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
OpenJudge Version
0.2.5, main
5d737c9.Description
The default strict Jaccard grader gives 0.0 for the same tool call when JSON argument keys are reordered:
{"city":"杭州","units":"metric"}versus{"units":"metric","city":"杭州"}. It compares Python dictionary representations, which preserve insertion order.Serialize arguments with sorted JSON keys on both sides. This also handles nested objects and objects inside arrays, while preserving array order, different values, tool names and existing set semantics. Update the matching documentation.
Validation:
aevaluateregressions on unchanged main: 6 failed, 29 passed.pre-commit run --all-files --verbose: all hooks pass, including configured pytest 1253 passed, 139 skipped (five existing NumPy zero-variance warnings).git diff --check: passed.Both JSON-text and dictionary argument inputs run through the real public grader. No model calls, services or replaced implementations are involved. The configured test hook excludes multimodal/generator/data/examples; skipped provider tests were not executed. Existing #235 changes one-to-one step matching, not this Jaccard path.
AI assistance: OpenAI Codex prepared the change and ran validation.
Checklist
pre-commit run --all-files.