Skip to content

diag(shell): 失败信息自动附带判读 (#73) - #287

Merged
JavaZeroo merged 1 commit into
mainfrom
diag/auto-summary
Oct 31, 2026
Merged

JavaZeroo merged 1 commit into
mainfrom
diag/auto-summary

Conversation

@JavaZeroo

Copy link
Copy Markdown
Owner

内容

run.py 在抛 shell probe failed 时,自动用 #286 的解析器把转储读回来,并把判读追加到异常信息:

shell probe failed: workspace-failure mode=ui-copy-system stage=7 what=…
  tab bottom:0 …
  tab main:0 … | bottom:0 rejected=0 drained=yes child=259; main:0 rejected=0 drained=yes child=259

为什么

此前每次失败都要人工从 CI 日志里找 workspace-failure、抄计数器、再自己判断 output_bytes == consumed;我在同一份 artifact 上就因为手工统计口径不一致连续更正过两次。

现在判读由代码产生,且与 #286 的测试共用同一实现,口径不会因人因轮次而变。

验证

  • experiments.shell.test_summarize_failure → 4 tests OK;
  • experiments.shell.test_shell → 39 tests,1 error 为缺 shell_queue_probe.exe(本机未构建该目标),与本次改动无关。

边界

  • 只在失败分支追加判读,不改期限与判定;
  • 无转储(例如非 Workspace 场景失败)时不追加任何内容。

Refs #73 #38

…73

The raised message now ends with the same drained/rejected/child verdict the summariser computes, so reading a failure no longer depends on manual counting.
@JavaZeroo
JavaZeroo merged commit 4635418 into main Oct 31, 2026
18 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant