flaky: atlas-host pty::tests::a_title_* fail on loaded Linux CI runners

未关闭
#549 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
3/5
预计耗时
1-2 天
新手友好度
72/100
Issue 类型
缺陷
描述清晰度
基本清楚
活跃度
活跃
技术栈
rust
领域
cli, testing

调研方向

Start with crates/atlas-host/src/pty/tests.rs, especially the two failing title tests around lines 545 and 569, and inspect how they wait for the shell to exit and collect pty_title events. Run these tests repeatedly under load to reproduce the timing failure. Done when the tests reliably observe the expected title events with a deadline and pass in the Linux CI run.

由索引模型根据 Issue 内容生成。

描述

area: terminal os: linux type: bug

atlas-host pty::tests::a_title_* fail intermittently on the Linux CI runner (Rust (Linux)), unrelated to the PR under test. A rerun of the failed job passes.

Evidence

  • Run 44101988098 (PR #461, head eb47a231), job 134104287831: FAIL atlas-host pty::tests::a_title_that_returns_to_an_earlier_value_is_a_change, at crates/atlas-host/src/pty/tests.rs:569:
    assertion left == right failed left: ["x"] right: ["x", "y", "x"]. The rerun of the same head passed.
  • Run 44104330309 (PR #487, head 2d01391f), job 134114260709 (Linux): FAIL atlas-host pty::tests::a_title_repeated_by_the_program_is_pushed_once at tests.rs:545. The rerun passed.
  • Frequency: 2 failures in the roughly 8 Rust (Linux) attempts I watched on 2026-11-17, while the runners were heavily queued (many PRs at once). Both ran in the full nextest run (4171 tests, 4 threads), not alone.

Suspected cause (not verified)

Both tests drive a shell that prints OSC titles 20 ms apart (sleep 0.02) and rely on each landing in its own 8 ms coalescing window of the pty reader, then call until_exited and read pty_title events with try_recv. On a loaded runner a 20 ms sleep can stretch, or the reader can drain two OSCs in one window, or the exit can be observed before the last pushes are flushed; the failure shows only the first title arriving (["x"]). The tests wait for the process to exit, not for the expected number of titles. Waiting for the expected events (with a deadline) instead of a fixed sleep and a post-exit drain would remove the timing dependency.

No fix in the PRs where it was seen.

主要语言
Rust
星标
3
派生
1
平均合并
9 小时 51 分钟
30 天内合并 PR
310

环境准备

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

codeatlasdev/atlas 的其他 Issue

查看 codeatlasdev/atlas 的全部 Issue

相似的 Issue

更多 Rust Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。