REGRESSION(262844@main): css3/filters/reference-filter-change-repaint.html is randomly failing

[cdumez]

2e76ca3

REGRESSION(262844@main): css3/filters/reference-filter-change-repaint.html is randomly failing
https://bugs.webkit.org/show_bug.cgi?id=255383

Reviewed by Said Abou-Hallawa.

262844@main slightly changed the ordering of attribute processing in SVGFETurbulenceElement
and it was caused this repaint test to become flaky. We need to make sure that
SVGFETurbulenceElement::attributeChanged() calls this first:
```
m_seed->setBaseValInternal(newValue.toFloat());
```
when the seed attribute changes, before calling the base class's attributeChanged.

This patch restores the ordering to address the issue.

* Source/WebCore/svg/SVGFETurbulenceElement.cpp:
(WebCore::SVGFETurbulenceElement::attributeChanged):

Canonical link: https://commits.webkit.org/262985@main

0da5a72

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🛠 gtk
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🧪 gtk-wk2
	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🧪 api-gtk
	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2
	✅ 🛠 tv-sim	✅ 🧪 mac-wk2-stress
	✅ 🛠 watch
✅ 🛠 🧪 unsafe-merge	✅ 🛠 watch-sim

[webkit-early-warning-system]

EWS run on current version of this PR (hash 0da5a72)

Details

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🛠 gtk
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🧪 gtk-wk2
	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🧪 api-gtk
	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2
	✅ 🛠 tv-sim	✅ 🧪 mac-wk2-stress
	✅ 🛠 watch
✅ 🛠 🧪 unsafe-merge	✅ 🛠 watch-sim

[cdumez]

Patch is ready for review. It fixes a very recent regression.

[shallawa]

The change seems fine. But how can we make sure 262844@main did not change the order of processing the attributes of something else? Or is this change going to affect layout tests only?

[cdumez]

The change seems fine. But how can we make sure 262844@main did not change the order of processing the attributes of something else? Or is this change going to affect layout tests only?

I tried to be careful when I made 262844@main but clearly I got the ordering slightly wrong in some cases. I think this is mostly an issue for SVG because of svgAttributeChanged(). I'll take another look at the SVG changes to double check ordering. I am also curious if we really need a separate svgAttributeChanged() for SVG elements.

[webkit-commit-queue]

Committed 262985@main (2e76ca3): https://commits.webkit.org/262985@main

Reviewed commits have been landed. Closing PR #12699 and removing active labels.