296134 2025-07-17 10:39:56 -0700 REGRESSION(296337@main): currentcolor used in custom property no longer resolves correctly 2026-03-08 14:53:45 -0700 1 1 1 Unclassified WebKit New Bugs Safari 18 Unspecified Unspecified RESOLVED FIXED https://github.com/web-platform-tests/wpt/pull/58355 InRadar P2 Normal --- 1 sam sam benoit.rouleau webkit-bug-importer oldest_to_newest 2130878 0 476073 sam 2025-07-17 10:39:56 -0700 Created attachment 476073 Test Case Benoît Rouleau (@benface) pointed out in https://github.com/w3c/csswg-drafts/issues/10371#issuecomment-3082053452 that we have a regression from 296337@main where `currentcolor` is no longer resolved correctly for rendering when used in a custom property: https://codepen.io/benface/pen/KwddNvX (Further simplified test case attached) 2130879 1 sam 2025-07-17 10:46:21 -0700 The issue is that during style building, rather than using the parsed content in the `Style::CustomProperty`, which has the correct "currentcolor" Style::Color in it, what we do is fall into a code path that calls, `Style::CustomProperty::tokens()`, and then re-parses the color. `CustomProperty::tokens()` uses `CustomProperty::propertyValueSerialization()`, but with no RenderStyle available, we end up serializing to black by default. Ideally we would just use the already parsed value in the CustomProperty, but I don't know enough about this code to understand why we don't. So, barring that, I can see two solution: 1. Store the original tokens in the CustomProperty at creation so that we have exactly what is needed. 2. Add additional serialization logic needed for this case back. I think 1 would less error prone. 2130880 2 sam 2025-07-17 10:47:02 -0700 I don't think this would be a trivial direct revert, as there has been a bit that has built on the serialization changes this provided. 2130888 3 benoit.rouleau 2025-07-17 11:10:23 -0700 Thank you for filing this! 🙏 This bug is breaking this website in production: https://thegraph.com. Notice the icons in the header are now black instead of white. 2130890 4 webkit-bug-importer 2025-07-17 11:14:06 -0700 <rdar://problem/156076502> 2130899 5 sam 2025-07-17 11:40:22 -0700 Ok, given this is a real world regression, going to try to do a partial revert, and just do some name changing to make things work. 2130925 6 sam 2025-07-17 12:38:36 -0700 Pull request: https://github.com/WebKit/WebKit/pull/48197 2131184 7 ews-feeder 2025-07-18 10:08:21 -0700 Committed 297600@main (6496fbaca29d): <https://commits.webkit.org/297600@main> Reviewed commits have been landed. Closing PR #48197 and removing active labels. 2188234 8 ahmad.saleem792 2026-03-08 14:53:45 -0700 Submitted web-platform-tests pull request: https://github.com/web-platform-tests/wpt/pull/58355 476073 2025-07-17 10:39:56 -0700 2025-07-17 10:39:56 -0700 Test Case test-custom-color-using-current-color.html text/html 292 sam PHN0eWxlPgogICAgQHByb3BlcnR5IC0tY29sb3IgewogICAgICBzeW50YXg6ICI8Y29sb3I+IjsK ICAgICAgaW5oZXJpdHM6IGZhbHNlOwogICAgICBpbml0aWFsLXZhbHVlOiBjdXJyZW50Y29sb3I7 CiAgICB9CgogICAgZGl2IHsKICAgICAgY29sb3I6IHJlZDsKICAgIH0KCiAgICBzcGFuIHsKICAg ICAgYmFja2dyb3VuZC1jb2xvcjogdmFyKC0tY29sb3IpOwogICAgfQo8L3N0eWxlPgo8cD5CbG9j ayBiZWxvdyBzaG91bGQgYmUgYWxsIHJlZDwvcD4KPGRpdj4KICA8c3Bhbj5oZWxsbzwvc3Bhbj4K PC9kaXY+Cg==