Skip to content

[Tracker] Terminal rendering corruption & visual glitches (stale-issue sweep, July 2026) #5605

Closed as duplicate

Description

@Kitenite

Consolidated tracker for stale terminal-rendering issues closed in the July 2026 stale-issue sweep (no activity since 2026-06-01).

Each issue below was closed as a duplicate of this tracker. If any of them still reproduces (or is still wanted) on a current build, comment here or on the original issue and we'll reopen it.

Activity

  1. github-actions commented on Jul 11, 2026

    @github-actions
    Contributor

    Automated triage note

    This is a consolidated tracker, not a single reproducible defect — it bundles 26 previously-closed terminal-rendering issues from the July 2026 stale sweep. I walked the desktop terminal rendering + transport subsystem (apps/desktop/src/renderer/lib/terminal/*, .../TabsContent/Terminal/*, apps/desktop/src/main/lib/terminal*) to check whether the cluster shares a single root cause that could be pinned with a bun:test reproduction. It doesn't, and a headless unit test can't exercise any of these — they only manifest against a real Electron renderer with WebGL/Canvas, a running font loader, and live flex layout:

    The pure, headless-testable helpers along this path (write-coalescer, the terminal-ws-transport framing, truncateUtf8ToLastBytes, convertLinkRangeToBuffer) are already byte-/boundary-safe and carry regression tests, and the subsystem already ships targeted mitigations for several of these exact reports — e.g. terminal-runtime.ts activates Unicode-11 widths before buffer replay citing #3572, and font-settle.ts, the write-coalescer, and the clipboard/appearance guards address the same families. So the tractable logic layer is already covered; what remains is genuinely renderer/GPU behavior.

    Recommendation: these need an Electron integration/E2E harness (drive the renderer over CDP, per apps/desktop/AGENTS.md) or manual verification on a current build, reproduced per individual issue — not one shared unit test. Per this tracker's own note, anyone still hitting a specific glitch on a current build should comment with the build version and repro steps so the individual issue can be reopened and triaged with the right harness.

    No PR opened — there is no single-root-cause defect to fix or to reproduce headlessly.

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

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions