docs: flake-vs-regression triage + git show cross-branch reads#101
Conversation
Dependency Review✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.Scanned FilesNone |
There was a problem hiding this comment.
Code Review
This pull request updates the documentation for git workflows and merge gate monitoring. It adds a guide on reading files from the object store using 'git show' and provides criteria for distinguishing between infrastructure flakes and real regressions in sharded test suites. The review comments suggest clarifying the scope of 'git diff' for file comparisons and refining the definition of 'real regressions' to better accommodate single-shard Playwright failures.
Important
The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.
…tion Gemini review on #101: `git diff <A> <B> -- <path>` compares a file, not the whole tree — reword and add `-- <path>` to the three-dot example. And the "real regression fails across all shards" line tensioned with the single-shard Playwright case — mark it the typical case, with the Playwright exception. Signed-off-by: Sebastian Mendel <info@sebastianmendel.de>
…tion Gemini review on #101: `git diff <A> <B> -- <path>` compares a file, not the whole tree — reword and add `-- <path>` to the three-dot example. And the "real regression fails across all shards" line tensioned with the single-shard Playwright case — mark it the typical case, with the Playwright exception. Signed-off-by: Sebastian Mendel <info@sebastianmendel.de>
3679e6e to
32d6147
Compare
…ession A /retro session (2026-07-16) found a red `E2E (1/4)` dismissed as a boot flake twice — correct once, but the second run was a real regression. Add a section on reading the shard's first error: a stack-boot/health-check line at the top means collateral failures below (infra flake, rerun); the same specs failing across all shards means a real bug. Notes the Playwright `locator.check: Test timeout` = actionability tell for a broken hit target. Signed-off-by: Sebastian Mendel <info@sebastianmendel.de>
Add `git show <branch>:<path>` for inspecting a file's committed content on another ref without checkout. A /retro session (2026-07-16) hit a phantom "file was modified" after a branch switch — the on-disk copy reflected the landed branch, not the one being reasoned about. Reading from the object store is authoritative. Signed-off-by: Sebastian Mendel <info@sebastianmendel.de>
…tion Gemini review on #101: `git diff <A> <B> -- <path>` compares a file, not the whole tree — reword and add `-- <path>` to the three-dot example. And the "real regression fails across all shards" line tensioned with the single-shard Playwright case — mark it the typical case, with the Playwright exception. Signed-off-by: Sebastian Mendel <info@sebastianmendel.de>
32d6147 to
824f8d8
Compare
|



Summary
Two reference additions from a
/retrosession: how to tell a one-shard-red CI failure apart (flake vs. real regression), and reading a file as committed on another branch without checkout.Came from
/retrosession on 2026-07-16 (timetracker release/deploy work).E2E (1/4)was dismissed as a boot-flake twice; the second run was a real regression a.field-check-rowrestyle had introduced (a spec'sgetByText-anchored.check()hung 30s). Separately, a phantom "file was modified" reminder after a branch switch nearly sent me chasing a non-edit.git checkoutinstead of the object store.locator.checktimeout = actionability regression). Usegit show <branch>:<path>to read committed state across branches.Change
merge-gate-watcher.md: new "One shard red: flake vs. real regression" section.advanced-git.md: new "Reading a File As It Is Committed on Another Branch" section.Test plan
Build/Scripts/validate-skill.sh— 0 errors, 0 warningsgit show <ref>:<path>,git diff A B -- path,gh run rerun <id> --failed)