Skip to content

Streaming replies re-parse only their last markdown block #1753

Closed

Description

@currentbits

src/components/Markdown.tsx re-parses the whole reply on every throttled update, so long replies get slower to render and the newest text lags. Split streamed text at top-level block boundaries and memoize each finished block, so only the last block re-parses. Fall back to whole-text rendering for syntax that links blocks together (reference links, footnotes, open code fences).

Activity

  1. added and removed on Oct 30, 2026
  2. currentbits commented on Oct 30, 2026

    @currentbits
    OwnerAuthor

    Already on main: since #1475, MarkdownBody splits a streaming reply with markdownChunks and only the last (live) chunk re-parses. Settled chunks are memoized and cached (src/components/Markdown.tsx). Nothing to build. This was filed from a check against a stale checkout.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions