aglinxinyuan commented on PR #7623:
URL: https://github.com/apache/texera/pull/7623#issuecomment-5338318161

   Conflict resolved, but it rescoped the PR — flagging that rather than 
burying it in a merge commit.
   
   **#7717 landed the jsdom half two days ago** (`testTimeout: 20000`, 
`hookTimeout: 30000`), which is exactly what conflicted. My branch carried 
30s/30s, so keeping mine would have reverted a value @mengw15 had just merged, 
with no argument for the change and inside a merge. I took main's verbatim 
instead — `git diff main -- frontend/vitest.config.ts` is now empty. If 20s 
turns out too tight for the test ceiling, that's a separate PR against a merged 
number, not a conflict resolution.
   
   Two statements were true before #7717 and are not now, so both got rewritten:
   
   - `vitest.browser.config.ts` justified `testTimeout: 30_000` as "the same 
30s the jsdom config uses". jsdom is 20s. The standing argument is that browser 
mode's 15s default is now the *tightest* per-test ceiling in the job, sitting 
on the slower of its two runtimes.
   - `TESTING.md` read "30s per test in both configs" under one shared 
provenance. Split per config against the post-#7717 values.
   
   **Separately, I got the job cap wrong and it's now 45, not 30.** I sized 30 
off run `31869822969` alone — legs at 8.6 / 10.0 / 10.7 min — and wrote "~3x 
the slowest". Pulling every frontend leg from the last eight `Required Checks` 
runs via the jobs API:
   
   | Leg | green legs observed | worst green |
   | --- | --- | --- |
   | ubuntu-latest | 8.4 – 9.6 min | 9.6 min |
   | windows-latest | 7.1 – 11.4 min | 11.4 min |
   | macos-latest | 10.0 – 21.1 min | **21.1 min** |
   
   The 21.1-minute one is run `32219352076` — this branch's own merge push, 
green. #7713 has the same range independently at 16m52s. So 30 was 1.4x the 
worst observed *passing* leg and could have ended a run that was only slow, 
which is the same false red the `fail-fast` opt-out is here to stop 
manufacturing. Per step, the variance is in `Prod build` (4.5 → 7.8 → 11.1 min 
on macOS) rather than the tests (4.4 → 5.3 min), so it's headroom for an 
Angular production build on a loaded runner, not for vitest.
   
   Title and description updated to match — the PR is now `fail-fast: false` + 
the job cap + the browser-mode ceiling + the doc row, and `Closes #6073` still 
holds since #7717 took the jsdom follow-up and this takes the other two.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to