2090202 - Some interop tests are failing because they're fuzzy.

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, defect, P2)

Description

[Emilio Cobos Álvarez [:emilio]]

Bug 2088007 made some tests fuzzy in our automation by touching the .ini files. They show as failures on wpt.fyi which means this impacts the interop score tho.

Can we move the fuzziness range to the test itself?

Comment 1

[Emilio Cobos Álvarez [:emilio]]

clip-path-shape-007.html, clip-path-shape-008.html, clip-path-shape-009.html, & clip-path-shape-010.html are the relevant ones.

Comment 2

[BugBot [:suhaib / :marco]]

Set release status flags based on info from the regressing bug 2088007

Comment 3

[Daniel Holbert [:dholbert]]

wpt.fyi link for convenience:
https://wpt.fyi/results/css/css-masking/clip-path?label=master&label=experimental&aligned&q=clip-path-shape-0

On wpt.fyi, all four tests seem to fail with the following metrics:

maxDifference: 73
totalPixels: 606

(though in our .ini files, we've got totalPixels marked as 640, so maybe our CI has slightly more pixels being fuzzy for some reason. shrug)

Looking at the failure message, it seems there's a ~two-pixel-thick band around the perimeter of the shape, which has different antialiasing applied (different darknesses of green).

These 4 tests don't currently have any fuzzy annotation in the test at all, so other browsers apparently don't need any fuzzy annotations.

It looks like the reason for the fuzziness is that these 4 tests use the clip-path property whereas their reference case uses an SVG clipPath element, and I guess we give those clipping techniques different treatment now, as of bug 2088007.

Comment 4

[Emilio Cobos Álvarez [:emilio]]

Attachment: Bug 2090202 - Support simple svg clipPath elements in CreateSimpleClipRegion. r=gw,nical,#gfx-reviewers

This along with bug 2090316 makes these tests not fuzzy, but we need
gfx.webrender.svg-simple-clips=true too.

So no behavior change for now.

Comment 5

[Emilio Cobos Álvarez [:emilio]]

Attachment: Bug 2090202 - Turn on webrender.svg-simple-clips. r=nical,gw,#gfx-reviewers

Comment 6

[Emilio Cobos Álvarez [:emilio]]

So I think comment 4 is the right approach, but it'd need gfx.webrender.svg-simple-clips=true as well to get the right coordinates for the clip...

Comment 7

[Daniel Holbert [:dholbert]]

(In reply to Emilio Cobos Álvarez [:emilio] from comment #6)

it'd need gfx.webrender.svg-simple-clips=true as well to get the right coordinates for the clip...

That pref was added (off-by-default) in bug 2070993 / https://phabricator.services.mozilla.com/D325279 .

nical, do we have a plan to get that pref enabled by default (and riding the trains) in the near term?

If there's not a clear path, then maybe we should proactively label the tests as fuzzy (in the tests themselves) as suggested in comment 0, to avoid taking an interop-2026 hit (with the score being finalized in the coming months).

(Given that we're the only ones to need the fuzz and we won't need it forever, perhaps it'd be worth including an explanatory HTML-comment alongside the fuzz annotation, saying that Firefox antialiases edges a bit differently betweenclip-path and SVG clipPath, with a bug link to where we intend to remove that difference, to increase the likelihood that the fuzz gets removed when we no longer need it.)

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/8493ed2cc78f
https://hg.mozilla.org/integration/autoland/rev/b37525ba1f95
Support simple svg clipPath elements in CreateSimpleClipRegion. r=firefox-svg-reviewers,layout-reviewers,jwatt

Comment 9

[Daniel Holbert [:dholbert]]

We probably need a new bug to track the fact that these WPTs still fail, right?

Comment 10

[Emilio Cobos Álvarez [:emilio]]

Not anymore with https://phabricator.services.mozilla.com/D340535. Sorry, I forgot to push it :)

Comment 11

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/3d747f45151a
https://hg.mozilla.org/integration/autoland/rev/971e2e10f292
Turn on webrender.svg-simple-clips. r=layout-reviewers,jwatt

Comment 12

[Daniel Holbert [:dholbert]]

(In reply to Daniel Holbert [:dholbert] from comment #7)

nical, do we have a plan to get that pref enabled by default (and riding the trains) in the near term?

Whoops, I didn't see comment 5's patch when posting this^ question. :)

I guess we're good with that landed now, yeah.

Comment 13

[Iulian Moraru]

https://hg.mozilla.org/mozilla-central/rev/b37525ba1f95
https://hg.mozilla.org/mozilla-central/rev/971e2e10f292

Comment 14

[BugBot [:suhaib / :marco]]

The patch landed in nightly and beta is affected.

:emilio, please make an uplift decision for beta:

• For beta, nominate the patch for beta uplift approval if the fix should be included in this release, or set status-firefox161 to wontfix if the fix can wait for the next release.

Uplift requests can be submitted from Lando: D340531, D340535.

See Requesting an Uplift for documentation on how to request an uplift.

For more information, please visit BugBot documentation.

Comment 15

[Emilio Cobos Álvarez [:emilio]]

Probably wontfix.