Stream emulated tool-call prose incrementally - #955
Conversation
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (3)
📝 WalkthroughWalkthroughPartial emulated tool-call parsing now uses dedicated text preparation during streaming. New tests cover truncated prose, incomplete markers, and marker-like text inside incomplete thinking blocks. ChangesPartial emulated tool-call parsing
Estimated code review effort: 3 (Moderate) | ~20 minutes Possibly related PRs
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
|
This pull request is currently a draft. Reviews will not take place until the PR is marked as ready for review. |
What changed
TOOL_CALLmarker prefix<think>blocksWhy
The previous partial parser discarded the entire final line. Models commonly generate answers without newlines, so tool-enabled streaming requests that did not call a tool could produce no content until completion.
Impact
Normal prose regains incremental TTFT while incomplete emulated tool calls remain hidden from client-visible text.
Validation
cargo test -p skippy-server --lib— 204 passedcargo clippy -p skippy-server --all-targets -- -D warningsSummary by CodeRabbit