Skip to content

RAL-183: browser tests forward only real crash lines, marked as real - #270

Merged
Castor-o7 merged 3 commits into
mainfrom
ral-183/real-lines
Nov 14, 2026
Merged

Castor-o7 merged 3 commits into
mainfrom
ral-183/real-lines

Conversation

@Castor-o7

@Castor-o7 Castor-o7 commented Nov 14, 2026 •

Copy link
Copy Markdown
Owner

RAL-183, assigned by the watchman, after the lamplighter read RAL-179's own faked teardown crash in ci-local's log (61d0dca check.log 738–739) as a real one.

  • Nothing from fakes: run_browser writes nothing to the test's own stderr from a run whose Chrome command isn't the real RALLY_TEST_CHROME, or that uses the stand-in coredumpctl. The tests that fake a crash still assert on the script's output. This covers RAL-160's forwarded retry as well, whose fake-zypak lines reached the log the same way.
  • Real lines are marked: what a real run forwards begins browser test: [real Chrome, real coredumpctl]. A forwarded retry says "crashed" or "aborted" by its status.
  • The stand-in core can't pass for real: its time is 1970 and its pid is 999999.
  • A core is tied to its run by time: a core within 2 s of the run's own crash at start is said as that crash's, because a crash at start is gone before any pid of it can be taken. The lamplighter's real abort at 12:05:52Z was listed by two runs, each as "not one of this run's". The segfault test checks this with a stand-in coredumpctl that lists the core at the stand-in Chrome's crash time.
  • Checked: the full suite passes, and no faked line appears in the test output.

🤖 Generated with Claude Code

Castor-o7 and others added 3 commits November 14, 2026 06:46
The lamplighter read RAL-179's own faked teardown crash in ci-local's log as a real one. run_browser
now forwards nothing from a run with a stand-in Chrome command or coredumpctl; a real run's lines
begin "browser test: [real Chrome, real coredumpctl]"; the stand-in core's time and pid are
impossible ones.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…sh's, by time

The lamplighter's real abort of 12:05:52Z was listed by two runs, each as not theirs: a crash at
start is gone before a pid of it can be taken. A core within 2 s of the run's own crash at start is
now said as that one's.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@Castor-o7
Castor-o7 merged commit bfca754 into main Nov 14, 2026
5 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