[Webkit-unassigned] [Bug 218716] New: [GLIB] mathml/presentation/non-bmp-operators-stretching.html is image diff failing in the first run

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=218716

            Bug ID: 218716
           Summary: [GLIB]
                    mathml/presentation/non-bmp-operators-stretching.html
                    is image diff failing in the first run
           Product: WebKit
           Version: WebKit Nightly Build
          Hardware: Unspecified
                OS: Unspecified
            Status: NEW
          Severity: Normal
          Priority: P2
         Component: MathML
          Assignee: webkit-unassigned at lists.webkit.org
          Reporter: lmoura at igalia.com

Created attachment 413601

  --> https://bugs.webkit.org/attachment.cgi?id=413601&action=review

Imgdiff of first failed run

mathml/presentation/non-bmp-operators-stretching.html

The test operators are not stretched in the prefix form when it is the first of its kind run, likely missing some kind of initialization. For example, running 3 iterations, the first one fails and the other 2 work fine. But with `--run-singly`, it fails in all 3 iterations.

Running the non-bmp-operators-spacing test before it also makes the test pass.

A similar behavior happens in MiniBrowser:

- Load mathml/presentation/non-bmp-operators-stretching.html. Not stretched
- Load mathml/presentation/non-bmp-operators-stretching-expected.html (or the spacing version)
- Navigate back to mathml/presentation/non-bmp-operators-stretching.html - Stretched
- A full reload brings back the failure.

I noticed this while investigating the failure in a WPE build[1], likely due to being run after a worker crash.

[1] https://build.webkit.org/results/WPE-Linux-64-bit-Release-Tests/r269580%20(20968)/results.html

Attached is the img diff of the failure. Some letters diff but only in its vertical position, due to the operator height.

As this is kinda corner case, I'll not update the expectations to avoid hiding eventual more reproducible failures.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20201109/56f11e27/attachment.htm>