Make the quote bar and the e2e suite survive a loaded runner #351

Merged

robbertbos merged 1 commit from ci-flakes-quotebar-and-e2e-timeout into main 2026-09-05 11:25:44 +00:00

Conversation 0 Commits 1 Files changed 4 +19 -6

robbertbos commented 2026-09-05 11:13:31 +00:00

Owner

Copy link

Every open PR is red, and none of them is at fault: PR #346's entire diff is a postgres digest in compose.dev.yaml, and it fails frontend-test and e2e like the rest. Two causes, both in the tests.

quoteBar.test.ts (frontend-test, 6 PRs). quoteLineStarts read syntaxTree(state) straight. LanguageState.init parses under a 20ms budget capped at the first 3000 characters and then calls takeTree(), so a state hands out whatever was finished. In an EditorView the parse worker catches up; a bare state in a test has no worker, so on a loaded runner the first EditorState.create in the file - the one that also pays the lezer grammar warm-up - came back with an empty tree and the assertion saw []. That is why only the first test in the file went red. ensureSyntaxTree now covers the range actually asked about, which fixes the app too: a quote past character 3000 got no bar until the next edit. The added case reproduces it deterministically with a document longer than the initial viewport.

e2e (all 10 PRs). Tests that run in one or two seconds locally were timing out at 33-35s, and Chromium died mid-test with Protocol error (Runtime.callFunctionOn): Internal server error, session closed. One push fans out to thirteen jobs onto a runner of capacity two; in the same run backend-test took 21m23s against its usual ~90s. So: two Playwright workers on CI instead of three, and a 60s test timeout there. Local runs keep three workers and 30s.

Verified locally: 1421 vitest tests green, 122 e2e green (59.6s), pre-commit --all-files green.

Still open, not fixed here: thirteen jobs per push is the underlying pressure. Worth a look at whether test-build and security-scan need to run on every PR push.

Every open PR is red, and none of them is at fault: PR #346's entire diff is a postgres digest in `compose.dev.yaml`, and it fails `frontend-test` and `e2e` like the rest. Two causes, both in the tests. **`quoteBar.test.ts` (frontend-test, 6 PRs).** `quoteLineStarts` read `syntaxTree(state)` straight. `LanguageState.init` parses under a 20ms budget capped at the first 3000 characters and then calls `takeTree()`, so a state hands out whatever was finished. In an `EditorView` the parse worker catches up; a bare state in a test has no worker, so on a loaded runner the *first* `EditorState.create` in the file - the one that also pays the lezer grammar warm-up - came back with an empty tree and the assertion saw `[]`. That is why only the first test in the file went red. `ensureSyntaxTree` now covers the range actually asked about, which fixes the app too: a quote past character 3000 got no bar until the next edit. The added case reproduces it deterministically with a document longer than the initial viewport. **e2e (all 10 PRs).** Tests that run in one or two seconds locally were timing out at 33-35s, and Chromium died mid-test with `Protocol error (Runtime.callFunctionOn): Internal server error, session closed`. One push fans out to thirteen jobs onto a runner of capacity two; in the same run `backend-test` took 21m23s against its usual ~90s. So: two Playwright workers on CI instead of three, and a 60s test timeout there. Local runs keep three workers and 30s. Verified locally: 1421 vitest tests green, 122 e2e green (59.6s), `pre-commit --all-files` green. Still open, not fixed here: thirteen jobs per push is the underlying pressure. Worth a look at whether `test-build` and `security-scan` need to run on every PR push.

robbertbos force-pushed ci-flakes-quotebar-and-e2e-timeout from be51b72b49 to 49a632c19e 2026-09-05 11:17:24 +00:00 Compare

robbertbos merged commit 64526356b3 into main 2026-09-05 11:25:44 +00:00

robbertbos deleted branch ci-flakes-quotebar-and-e2e-timeout 2026-09-05 11:25:44 +00:00

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

bug

Something is not working

  

duplicate

This issue or pull request already exists

  

enhancement

New feature

  

help wanted

Need some help

  

invalid

Something is wrong

  

nldd ds

  

question

More information is needed

  

wontfix

This won't be fixed

No labels

bug

duplicate

enhancement

help wanted

invalid

nldd ds

question

wontfix

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

robbertbos/waggle!351

No description provided.