Fix Responses history observation after tool argument normalization - #962
Merged
Merged
Conversation
bradhilton
marked this pull request as ready for review
September 25, 2026 06:54
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Responses requests containing historical function calls can generate successfully and then fail in ART's history observer. vLLM renders JSON argument strings into mappings, while the observer reconstructs a typed ChatCompletionRequest that requires strings.
Reuse the existing OpenAI argument-normalization helper when constructing that observer request. The helper is moved unchanged; native requests, rendered history, generated outputs, and append-only prefix checks retain their existing behavior.
Validation: two causal streaming/non-streaming regressions fail before the fix; 42 focused inference-history and append-only tests pass afterward. Ruff and focused ty pass. Independent Astra and Fable source reviews are CLEAR for exact head 273a3d8; both review sessions and their source/test limitations are retained. Hosted Prek 36103708399 passed: 1,218 Megatron-lane cases; 1,928 unit cases with 34 explicit skips and 2 deselections; all statics passed. GPU validation classified this change as not requiring a GPU run. This PR does not deploy the fix or change any model/dependency pin.
Merged after existing image build 36100788410 completed successfully and its temporary resources were reconciled. Merge c462353 has the exact reviewed/tested tree. Automatic image build 36105650388 is separate from a serving deployment; no live policy success is claimed by this source fix.