[css-pseudo] Highlight pseudos and forced-color-adjust #7264

dandclark · 2022-05-10T00:07:40Z

I'd like to try to clarify how forced-color-adjust should apply to highlight pseudos.

In crbug.com/1309835, @andruud raised the question of how forced-color-adjust:preserve-parent-color should interact with the special definition of currentColor used with highlight inheritance. Normally preserve-parent-color has currentColor compute to the used value of the parent's color, but with highlights it's not so clear what this value should resolve to. Making it dependent on which highlight overlays are active could be problematic as that info may not be readily available until paint-time.

A simple way to resolve any strangeness could be to just treat forced-color-adjust:preserve-parent-color as forced-color-adjust:none for highlights. The motivating use case for preserve-parent-color involving SVG icons doesn't seem to be applicable to highlight pseudos as far as I can tell.

Considering things more broadly, does it make sense to allow forced-color-adjust in highlight pseudos at all? It is not one of the properties listed at https://drafts.csswg.org/css-pseudo-4/#highlight-styling, but Chrome allows it to be used with ::selection and the other highlight pseudos when --enable-blink-features=HighlightInheritance is enabled. Example here. (When HighlightInheritance is not enabled, Chrome looks like it always ignores forced colors for ::selection pseudos, which seems clearly wrong).

Perhaps highlights should always just use the forced-colors state of the originating element. Does it actually make sense to allow authors to opt out of forced colors for highlights but not the element containing the underlying text? Or vice versa?

cc @alisonmaher @delan @mrego

The text was updated successfully, but these errors were encountered:

delan · 2022-05-10T08:06:13Z

(ping @fantasai, @frivoal)

css-meeting-bot · 2022-05-18T16:36:45Z

The CSS Working Group just discussed [css-pseudo] Highlight pseudos and forced-color-adjust, and agreed to the following:

RESOLVED: forced-color-adjust doesn't do anything on highlight pseudos, and highlight pseudos take their forced color state from the originating element

The full IRC log of that discussion

<dbaron> Topic: [css-pseudo] Highlight pseudos and forced-color-adjust
<emilio> (I guess that's what chrishtr just said)
<dbaron> github: https://github.com//issues/7264
<dbaron> dandclark: question about how forced-color-adjust:preserve-parent-color should work with highlight inheritance
<dbaron> dandclark: this interaction is potentially problematic.
<fantasai> emilio, yeah, and it's obnoxious when I try to load a page and it gives me a blank forever even though I know the text is loaded
<dbaron> dandclark: Prereq: consider wether forced-color-adjust should be allowed in highlight pseudos.
<dbaron> dandclark: List of properties. Currently not allowed there.
<Rossen_> q?
<dbaron> s/Prereq/Prerequisite/
<dbaron> s/List/The spec gives a list/
<dbaron> s/wether/whether/
<dbaron> fantasai: I think the idea of having it not apply and just applying the forced-color-adjust effect of the originating element makes sense to me.
<dbaron> fantasai: I think that was mentioned in the post as a way to deal with this.
<dbaron> dandclark: I think that's a good outcome... seems simplest.
<dbaron> dandclark: For all other color based properties the highlights are sort of an independent cascade. But I think it makes sense here. Seems odd that the originating element would take forced colors and the highlights not.
<dbaron> dandclark: If the originating element has forced-color-adjust:preserve-parent-color, what does that mean for highlight pseudos that are active over that originating element?
<dbaron> fantasai: Effect is that the color property is able to inherit from the parent the actual color of the parent... and that doesn't really impact the highlight pseudos ... which doesn't matter for highlight pseudos where the currentcolor keyword comes from the previous layer. So the keyword wouldn't have an effect.
<dbaron> dandclark: so then try to resolve that forced-color-adjust doesn't do anything on highlight pseudos, and that highlight pseudos take forced color state of originating element
<dbaron> Rossen: Proposed resolution: forced-color-adjust doesn't do anything on highlight pseudos, and highlight pseudos take their forced color state from the originating element
<dbaron> RESOLVED: forced-color-adjust doesn't do anything on highlight pseudos, and highlight pseudos take their forced color state from the originating element

See official minutes

dandclark · 2022-05-18T20:13:05Z

@fantasai Could you help me understand what these tags mean?
I'm guessing that Needs Edits means that a spec change is needed to reflect the meeting resolution, but I'm not sure about Closed Accepted by CSSWG Resolution.

Sorry if these are documented somewhere; I looked but couldn't find anything.

…ement Per the resolution in [1], forced-color-adjust should not be one of the supported properties [2] in highlight pseudos, and the forced-color state of highlights should be taken from the originating element. This change updates the HighlightInheritance implementation to match the resolution. There are a few parts to this: - css_properties.json5 is updated so that forced-color-adjust is no longer a valid property for highlight pseudos. But, since shipping this change for ::selection may cause compatibility issues, we introduce a new valid_for_highlight_legacy parameter that maintains the old behavior and we update valid_for_highlight, which will be used only for highlights with the new inheritance model enabled, to use the new behavior. - This uncovered an older bug where DetermineValidPropertyFilter wasn't using any filter at all for ::highlight(), ::spelling-error, and ::grammar-error. This is now fixed. - StyleResolver::ApplyInheritance is updated to propagate forced-colors status from the originating element to the corresponding highlight pseudo. The new test is disabled on Mac/Linux due to some small differences in text decoration painting between those from highlight pseudos and from non-highlight rules, unrelated to the contents of this change. [1] w3c/csswg-drafts#7264 (comment) [2] https://drafts.csswg.org/css-pseudo-4/#highlight-styling Bug: 1309835, 1024156 Change-Id: I732afd24bbba2619a6718270faa0a1d2fcb98512

…ement Per the resolution in [1], forced-color-adjust should not be one of the supported properties [2] in highlight pseudos, and the forced-color state of highlights should be taken from the originating element. This change updates the HighlightInheritance implementation to match the resolution. There are a few parts to this: - css_properties.json5 is updated so that forced-color-adjust is no longer a valid property for highlight pseudos. But, since shipping this change for ::selection may cause compatibility issues, we introduce a new valid_for_highlight_legacy parameter that maintains the old behavior and we update valid_for_highlight, which will be used only for highlights with the new inheritance model enabled, to use the new behavior. - This uncovered an older bug where DetermineValidPropertyFilter wasn't using any filter at all for ::highlight(), ::spelling-error, and ::grammar-error. This is now fixed. - StyleResolver::ApplyInheritance is updated to propagate forced-colors status from the originating element to the corresponding highlight pseudo. The new test is disabled on Mac/Linux due to some small differences in text decoration painting between those from highlight pseudos and from non-highlight rules, unrelated to the contents of this change. [1] w3c/csswg-drafts#7264 (comment) [2] https://drafts.csswg.org/css-pseudo-4/#highlight-styling Bug: 1309835, 1024156 Change-Id: I732afd24bbba2619a6718270faa0a1d2fcb98512 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3665644 Reviewed-by: Rune Lillesveen <[email protected]> Reviewed-by: Delan Azabani <[email protected]> Commit-Queue: Dan Clark <[email protected]> Cr-Commit-Position: refs/heads/main@{#1009639}

…ement Per the resolution in [1], forced-color-adjust should not be one of the supported properties [2] in highlight pseudos, and the forced-color state of highlights should be taken from the originating element. This change updates the HighlightInheritance implementation to match the resolution. There are a few parts to this: - css_properties.json5 is updated so that forced-color-adjust is no longer a valid property for highlight pseudos. But, since shipping this change for ::selection may cause compatibility issues, we introduce a new valid_for_highlight_legacy parameter that maintains the old behavior and we update valid_for_highlight, which will be used only for highlights with the new inheritance model enabled, to use the new behavior. - This uncovered an older bug where DetermineValidPropertyFilter wasn't using any filter at all for ::highlight(), ::spelling-error, and ::grammar-error. This is now fixed. - StyleResolver::ApplyInheritance is updated to propagate forced-colors status from the originating element to the corresponding highlight pseudo. The new test is disabled on Mac/Linux due to some small differences in text decoration painting between those from highlight pseudos and from non-highlight rules, unrelated to the contents of this change. [1] w3c/csswg-drafts#7264 (comment) [2] https://drafts.csswg.org/css-pseudo-4/#highlight-styling Bug: 1309835, 1024156 Change-Id: I732afd24bbba2619a6718270faa0a1d2fcb98512 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3665644 Reviewed-by: Rune Lillesveen <[email protected]> Reviewed-by: Delan Azabani <[email protected]> Commit-Queue: Dan Clark <[email protected]> Cr-Commit-Position: refs/heads/main@{#1009639}

…tatus of their originating element, a=testonly Automatic update from web-platform-tests Highlight pseudos use the forced-color status of their originating element Per the resolution in [1], forced-color-adjust should not be one of the supported properties [2] in highlight pseudos, and the forced-color state of highlights should be taken from the originating element. This change updates the HighlightInheritance implementation to match the resolution. There are a few parts to this: - css_properties.json5 is updated so that forced-color-adjust is no longer a valid property for highlight pseudos. But, since shipping this change for ::selection may cause compatibility issues, we introduce a new valid_for_highlight_legacy parameter that maintains the old behavior and we update valid_for_highlight, which will be used only for highlights with the new inheritance model enabled, to use the new behavior. - This uncovered an older bug where DetermineValidPropertyFilter wasn't using any filter at all for ::highlight(), ::spelling-error, and ::grammar-error. This is now fixed. - StyleResolver::ApplyInheritance is updated to propagate forced-colors status from the originating element to the corresponding highlight pseudo. The new test is disabled on Mac/Linux due to some small differences in text decoration painting between those from highlight pseudos and from non-highlight rules, unrelated to the contents of this change. [1] w3c/csswg-drafts#7264 (comment) [2] https://drafts.csswg.org/css-pseudo-4/#highlight-styling Bug: 1309835, 1024156 Change-Id: I732afd24bbba2619a6718270faa0a1d2fcb98512 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3665644 Reviewed-by: Rune Lillesveen <[email protected]> Reviewed-by: Delan Azabani <[email protected]> Commit-Queue: Dan Clark <[email protected]> Cr-Commit-Position: refs/heads/main@{#1009639} -- wpt-commits: 23505765f04099bac1909b4a1400eb0c02829061 wpt-pr: 34223

…ement Per the resolution in [1], forced-color-adjust should not be one of the supported properties [2] in highlight pseudos, and the forced-color state of highlights should be taken from the originating element. This change updates the HighlightInheritance implementation to match the resolution. There are a few parts to this: - css_properties.json5 is updated so that forced-color-adjust is no longer a valid property for highlight pseudos. But, since shipping this change for ::selection may cause compatibility issues, we introduce a new valid_for_highlight_legacy parameter that maintains the old behavior and we update valid_for_highlight, which will be used only for highlights with the new inheritance model enabled, to use the new behavior. - This uncovered an older bug where DetermineValidPropertyFilter wasn't using any filter at all for ::highlight(), ::spelling-error, and ::grammar-error. This is now fixed. - StyleResolver::ApplyInheritance is updated to propagate forced-colors status from the originating element to the corresponding highlight pseudo. The new test is disabled on Mac/Linux due to some small differences in text decoration painting between those from highlight pseudos and from non-highlight rules, unrelated to the contents of this change. [1] w3c/csswg-drafts#7264 (comment) [2] https://drafts.csswg.org/css-pseudo-4/#highlight-styling Bug: 1309835, 1024156 Change-Id: I732afd24bbba2619a6718270faa0a1d2fcb98512 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3665644 Reviewed-by: Rune Lillesveen <[email protected]> Reviewed-by: Delan Azabani <[email protected]> Commit-Queue: Dan Clark <[email protected]> Cr-Commit-Position: refs/heads/main@{#1009639} NOKEYCHECK=True GitOrigin-RevId: e3dd4411efafbe9d4c285898e2c195c55612fe89

…#7264

fantasai · 2022-12-29T21:01:26Z

@dandclark The Needs Edits label, as you guessed, means that the spec needs edits to reflect the resolution; Closed Accepted by CSSWG Resolution means that the issue was resolved by a CSSWG decision. There's other Closed labels that reflect whether it was accepted or rejected, and what type of acceptance... these are currently only documented in https://github.com/w3c/csswg-drafts/blob/main/bin/issuegen.pl#L104 and I mainly use it as a forcing function to think about what kind of decision or review might be needed.

The resolution is now edit in as 05e463d ; please let me know if that reflects your understanding of the resolution of this issue! Thanks!

dandclark · 2023-01-05T18:40:32Z

Thanks for clarifying! 05e463d looks right to me, so I'll go ahead and fully Close this.

…w3c#7264

The updated tests relied on ‘forced-color-adjust’ being applicable to highlight pseudos, which is no longer the case as of: w3c/csswg-drafts#7264 Bug: 1400970 Change-Id: I0872a85903f2abf73edd50429be83a48c3bc735e

The updated tests relied on ‘forced-color-adjust’ being applicable to highlight pseudos, which is no longer the case as of: w3c/csswg-drafts#7264 Bug: 1400970 Change-Id: I0872a85903f2abf73edd50429be83a48c3bc735e Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4705311 Reviewed-by: Stephen Chenney <[email protected]> Commit-Queue: Stephen Chenney <[email protected]> Cr-Commit-Position: refs/heads/main@{#1175025}

The updated tests relied on ‘forced-color-adjust’ being applicable to highlight pseudos, which is no longer the case as of: w3c/csswg-drafts#7264 Bug: 1400970 Change-Id: I0872a85903f2abf73edd50429be83a48c3bc735e Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4705311 Reviewed-by: Stephen Chenney <[email protected]> Commit-Queue: Stephen Chenney <[email protected]> Cr-Commit-Position: refs/heads/main@{#1175025}

…t apply to pseudo-highlights, a=testonly Automatic update from web-platform-tests Update test: forced-color-adjust does not apply to pseudo-highlights The updated tests relied on ‘forced-color-adjust’ being applicable to highlight pseudos, which is no longer the case as of: w3c/csswg-drafts#7264 Bug: 1400970 Change-Id: I0872a85903f2abf73edd50429be83a48c3bc735e Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4705311 Reviewed-by: Stephen Chenney <[email protected]> Commit-Queue: Stephen Chenney <[email protected]> Cr-Commit-Position: refs/heads/main@{#1175025} -- wpt-commits: 2217dea6028932dbda4f7188ddd275cac80c82fb wpt-pr: 41140

The updated tests relied on ‘forced-color-adjust’ being applicable to highlight pseudos, which is no longer the case as of: w3c/csswg-drafts#7264 Bug: 1400970 Change-Id: I0872a85903f2abf73edd50429be83a48c3bc735e Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4705311 Reviewed-by: Stephen Chenney <[email protected]> Commit-Queue: Stephen Chenney <[email protected]> Cr-Commit-Position: refs/heads/main@{#1175025}

dandclark added the css-pseudo-4 Current Work label May 10, 2022

dandclark added the Agenda+ label May 16, 2022

css-meeting-bot removed the Agenda+ label May 18, 2022

mozilla-apprentice mentioned this issue May 18, 2022

[css-pseudo] Highlight pseudos and forced-color-adjust mozilla/wg-decisions#790

Open

fantasai added Agenda+ Closed Accepted by CSSWG Resolution Needs Edits labels May 18, 2022

astearns removed the Agenda+ label May 18, 2022

chromium-wpt-export-bot mentioned this issue May 26, 2022

Highlight pseudos use the forced-color status of their originating element web-platform-tests/wpt#34223

Merged

fantasai added a commit that referenced this issue Dec 29, 2022

[css-pseudo-4] Define forced colors mode on highlight pseudo-elements. …

05e463d

…#7264

fantasai removed the Needs Edits label Dec 29, 2022

fantasai added the Commenter Response Pending label Dec 29, 2022

dandclark closed this as completed Jan 5, 2023

jakearchibald pushed a commit to jakearchibald/csswg-drafts that referenced this issue Jan 16, 2023

[css-pseudo-4] Define forced colors mode on highlight pseudo-elements. …

e21feda

…w3c#7264

delan mentioned this issue Feb 8, 2023

[css-pseudo] Custom properties on :root #6641

Closed

delan mentioned this issue Jun 6, 2023

[css-pseudo] highlight pseudos and non-applicable properties #7591

Closed

spectranaut mentioned this issue Jul 12, 2023

Update forced-colors-mode as it applies to pseudo highlights web-platform-tests/wpt#40998

Closed

chromium-wpt-export-bot mentioned this issue Jul 21, 2023

Update test: forced-color-adjust does not apply to pseudo-highlights web-platform-tests/wpt#41140

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[css-pseudo] Highlight pseudos and forced-color-adjust #7264

[css-pseudo] Highlight pseudos and forced-color-adjust #7264

dandclark commented May 10, 2022

delan commented May 10, 2022

css-meeting-bot commented May 18, 2022 •

edited by fantasai

Loading

dandclark commented May 18, 2022

fantasai commented Dec 29, 2022

dandclark commented Jan 5, 2023

[css-pseudo] Highlight pseudos and forced-color-adjust #7264

[css-pseudo] Highlight pseudos and forced-color-adjust #7264

Comments

dandclark commented May 10, 2022

delan commented May 10, 2022

css-meeting-bot commented May 18, 2022 • edited by fantasai Loading

dandclark commented May 18, 2022

fantasai commented Dec 29, 2022

dandclark commented Jan 5, 2023

css-meeting-bot commented May 18, 2022 •

edited by fantasai

Loading