1973210 - New wpt failures in /css/css-view-transitions/fractional-translation-from-transform.html

Status: NEW

(Core :: CSS Parsing and Computation, defect)

Description

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Syncing wpt PR 53221 found new untriaged test failures in CI

Tests Affected

New Tests That Don't Pass

• /css/css-view-transitions/fractional-translation-from-transform.html [wpt.fyi]: FAIL (Chrome: PASS, Safari: FAIL)

CI Results

Gecko CI (Treeherder)
GitHub PR Head

Notes

These updates will be on mozilla-central once bug 1972697 lands.

Note: this bug is for tracking fixing the issues and is not
owned by the wpt sync bot.

This bug is linked to the relevant tests by an annotation in
https://github.com/web-platform-tests/wpt-metadata. These annotations
can be edited using the wpt interop dashboard
https://jgraham.github.io/wptdash/

If this bug is split into multiple bugs, please also update the
annotations, otherwise we are unable to track which wpt issues are
already triaged. Resolving as duplicate or closing this issue should
be cause the bot to automatically update or remove the annotation.

Comment 1

[Daniel Holbert [:dholbert]]

This might just be a fuzzy failure? The screenshots look basically identical...

Comment 2

[Daniel Holbert [:dholbert]]

Never mind -- it's fuzzy, but kinda egregiously so; basically every pixel is fuzzy, because we're rendering the "X" glyphs at a slightly different position.

The test has a fuzzy annotation:

<meta name="fuzzy" content="maxDifference=0-100; totalPixels=0-35">

but our fuzzy failure on wpt.fyi is much more substantial (slightly smaller maxDifference, but massive totalPixels):

maxDifference: 74
totalPixels: 6056 

If I view a local copy of the testcase, and manually run document.startViewTransition() in Web Console, I can see the content immediately jump slightly (particularly if I zoom in, which makes the difference a bit more pronounced). That movement is the same underlying issue as the test failure, I'm pretty sure. I don't see any such movement in Chrome on the same machine (or in Safari on another machine).

Comment 3

[Daniel Holbert [:dholbert]]

Attachment: reduced testcase 1 (does the text move slightly when you click the button?)

Here's a somewhat-reduced testcase. When you click the button (you can only do it once per pageload), the text looks like it moves slightly to the right, in Firefox Nightly on my Linux machine.

Comment 4

[Daniel Holbert [:dholbert]]

Attachment: screencast showing the subtle movement when you click the button in reduced testcase 1

Comment 5

[Daniel Holbert [:dholbert]]

(In reply to Daniel Holbert [:dholbert] from comment #1)

This might just be a fuzzy failure? The screenshots look basically identical...
(In reply to Daniel Holbert [:dholbert] from comment #2)
Never mind -- it's fuzzy, but kinda egregiously so; basically every pixel is fuzzy, because we're rendering the "X" glyphs at a slightly different position [...]
our fuzzy failure on wpt.fyi is much more substantial (slightly smaller maxDifference, but massive totalPixels):

maxDifference: 74
totalPixels: 6056 

I dug into the test history a bit, and I think the fuzzy-failure that we're hitting is in fact precisely the thing that this test is testing for. (So, it wouldn't make sense to adjust the test to try to avoid the fuzzy failure here.)

Here's one relevant comment from the bug that added this test, where khushalsagar mentions seeing fuzzy text resulting from fractional transforms (and I think that observation is what spawned this test): https://issues.chromium.org/issues/40252438#comment9

(That bug was originally filed for a worse issue, but it resulted in khushalsagar noticing this fuzzy-text issue and fixing & adding a test for it.)

Comment 6

[Emilio Cobos Álvarez [:emilio]]

The issue here seems to be that SnapRect (here) doesn't account for the final transformed rect.

It seems somewhat minor, so we could probably argue for removing this test, but fixing it also doesn't seem terrible? Somewhat painful tho.