Skip to content

Restore bounded wait in QuickAccessDialogTest to avoid bundle timeout#4096

Merged
vogella merged 1 commit into
eclipse-platform:masterfrom
vogella:lv/quickaccess-test-timeout-fix
Jun 15, 2026
Merged

Restore bounded wait in QuickAccessDialogTest to avoid bundle timeout#4096
vogella merged 1 commit into
eclipse-platform:masterfrom
vogella:lv/quickaccess-test-timeout-fix

Conversation

@vogella

@vogella vogella commented Jun 15, 2026

Copy link
Copy Markdown
Contributor

The dialog-initialization wait in testPreviousChoicesAvailableForExtension was set to TIMEOUT * 1000, roughly 83 minutes. When the dialog initialization stalls, which happens on the macOS CI runners, the wait blocks well past the 1200s tycho-surefire budget for the whole org.eclipse.ui.tests bundle, so the test runtime is killed with SIGTERM (exit code 143). Maven then reports a launch failure with no test results, which is why the macOS validation has been failing for every PR without pointing at a specific test.

Restoring the wait to TIMEOUT makes the test fail fast and locally when the dialog does not initialize, instead of hanging and taking the entire bundle down with it. The underlying race that leaves the proposals empty is addressed separately in #4090; this change only removes the accidental long wait so failures stay attributable and the bundle keeps reporting results. See #4009.

The wait for the dialog initialization in
testPreviousChoicesAvailableForExtension used TIMEOUT * 1000, about
83 minutes. When the dialog initialization stalls (notably on macOS),
this wait blocks far past the 1200s tycho-surefire budget for the whole
org.eclipse.ui.tests bundle, so the test runtime is killed with SIGTERM
(exit code 143) and no test results are reported for the bundle.

Restoring the wait to TIMEOUT lets the test fail fast and locally
instead of hanging and taking down the entire bundle.

See eclipse-platform#4009
@vogella

vogella commented Jun 15, 2026

Copy link
Copy Markdown
Contributor Author

@trancexpress the 83 min timeout which you places here, look like a typo to me and leads to process killing for the verification jobs. I suggest we change it back to the regular timeout to get a reasonable feedback (test errors) instead of a killed process.

@trancexpress

Copy link
Copy Markdown
Contributor

@vogella sorry! I must have committed a debug change without realizing it...

@github-actions

Copy link
Copy Markdown
Contributor

Test Results

   864 files  ±0     864 suites  ±0   53m 19s ⏱️ - 5m 35s
 8 050 tests ±0   7 807 ✅ +1  243 💤 ±0  0 ❌  - 1 
20 589 runs  ±0  19 934 ✅ +1  655 💤 ±0  0 ❌  - 1 

Results for commit c09b7f9. ± Comparison against base commit e314e02.

@vogella

vogella commented Jun 15, 2026

Copy link
Copy Markdown
Contributor Author

@vogella sorry! I must have committed a debug change without realizing it...

No worries, the timeout kill was just misleading for me, I assumed for a few days that we have an infrastructure problem.

@vogella vogella merged commit 4e89c0c into eclipse-platform:master Jun 15, 2026
18 checks passed
@vogella vogella deleted the lv/quickaccess-test-timeout-fix branch June 15, 2026 11:55
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.

2 participants