Fix mid-sentence message splits; recover stalled streams - #5
Merged
Conversation
…hdog
A stream interruption mid-turn makes Think persist the partial and
append the continuation as a NEW text part — often mid-sentence. The
chat renderer treated each text part as a standalone markdown document,
tearing sentences and lists apart at the seam ("Get a major" / "'s full
requirement tree…"). Consecutive text parts are now concatenated before
rendering.
Also sets chatStreamStallTimeoutMs = 120s so a model stream that parks
without erroring gets aborted into Think's bounded recovery path
instead of leaving the client spinning — the likely source of the
"spotty" feel.
Claude-Session: https://claude.ai/code/session_01MKLJUWk6biNAKXupHTTWn5
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Recovery continuations append new text parts mid-sentence; the renderer now coalesces consecutive text parts instead of rendering each as its own markdown block. Adds a 120s stream-stall watchdog so parked model streams route into Think's bounded recovery rather than spinning.
https://claude.ai/code/session_01MKLJUWk6biNAKXupHTTWn5