Bug 223973 - [ macOS Debug ] imported/w3c/web-platform-tests/xhr/xhr-timeout-longtask.any.worker.html is a flakey text failure
Summary: [ macOS Debug ] imported/w3c/web-platform-tests/xhr/xhr-timeout-longtask.any....
Status: NEW
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: WebKit Nightly Build
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Nobody
URL:
Keywords: InRadar
Depends on:
Blocks:
 
Reported: 2021-03-30 17:52 PDT by Robert Jenner
Modified: 2021-05-05 07:17 PDT (History)
3 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Robert Jenner 2021-03-30 17:52:05 PDT
imported/w3c/web-platform-tests/xhr/xhr-timeout-longtask.any.worker.html

is a flakey text failure in macOS Debug. 

Most recently, the failures have been more common in BigSur on Apple Silicon Macs. However, the test has been flakey for a while, and has occurred in BigSur Debug on Intel, as well as in Catalina egpu process. 

HISTORY URL:
https://results.webkit.org/?suite=layout-tests&test=imported%2Fw3c%2Fweb-platform-tests%2Fxhr%2Fxhr-timeout-longtask.any.worker.html&platform=mac

TEXT DIFF:
-PASS Long tasks should not trigger load timeout
+FAIL Long tasks should not trigger load timeout assert_unreached: [object XMLHttpRequestProgressEvent] Reached unreachable code
Comment 1 Robert Jenner 2021-03-30 17:52:53 PDT
Test is extremely flakey on Apple Silicon, and even more so on Intel. I did attempt to reproduce it, but was unable to on my Intel Mac. I would try on an Apple Silicon Mac, but I don't have access to such a machine.
Comment 2 Radar WebKit Bug Importer 2021-03-30 17:54:26 PDT
<rdar://problem/76030809>
Comment 3 Robert Jenner 2021-03-30 18:24:53 PDT
Updated test expectations to Pass Failure here:
https://trac.webkit.org/changeset/275252/webkit
Comment 4 youenn fablet 2021-05-05 07:17:52 PDT
It might be that the load does not complete in 100 ms as the debug bot is slow, there is only 50ms to do this before hitting the timeout.

We could try reducing load delay to 10 ms, but I am not sure 140ms will be sufficient anyway in debug.

I think we can either skip the test in Debug or leave it as Pass/Failure.
Pass/Failure is probably good enough.