2048804 - Regression involving revert-layer, linear-gradient, custom properties

Status: ASSIGNED

(Core :: CSS Parsing and Computation, defect)

Description

[Kirk L]

Attachment: regression.html

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:153.0) Gecko/20100101 Firefox/153.0

Steps to reproduce:

https://codepen.io/kirk-l/pen/PwWWVba?editors=1100

This unusual combination of styles is from me using revert-layer to style certain elements on a page with Tailwind CSS, and keep rest of the page's legacy styles from affecting this element.

Strangely, if I change the code any of these ways, the bug doesn't manifest:

• Don't use all: revert-layer.
• Inline the value of var(--tw-gradient-stops) instead of using a var.
• Use the fallback of var(--tw-gradient-via-stops) directly. (Note this var doesn't exist.)
• Inline the value of var(--tw-gradient-from) instead of using a var.
• Don't register the @property --tw-gradient-from.

The code works as expected on Chrome 149 and Safari 26.5.

Actual results:

One box shows a linear-gradient background, but the box with the style all: revert-layer; does not show the background. This happens in Firefox 153 and Nightly (154)

Expected results:

Two identical boxes should appear. This works on Firefox 152.

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

The Bugbug bot thinks this bug should belong to the 'Core::CSS Parsing and Computation' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 2

[Alice0775 White]

Rgression window:
https://hg-edge.mozilla.org/integration/autoland/pushloghtml?fromchange=f630dba53a17b5cb588464347209121b4c5040fb&tochange=a0268b86ab8fd3a85ef1adeb2203c64cb341b326

Comment 3

[BugBot [:suhaib / :marco/ :calixte]]

:emilio, since you are the author of the regressor, bug 2046992, could you take a look? Also, could you set the severity field?

For more information, please visit BugBot documentation.

Comment 4

[Emilio Cobos Álvarez [:emilio]]

Attachment: More reduced test-case

Thanks. So the issue seems to be that we fail to detect the dependency on the registered color via the property fallback.

Comment 5

[Emilio Cobos Álvarez [:emilio]]

Swarup is currently working on bug 2042898 which should fix this, but I'll try to land something more surgical to fix the regression.

Comment 6

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 2046992

Comment 7

[Pascal Chevrel (relman team) -> :pascalc]

Regressor backed out of beta

Comment 8

[Emilio Cobos Álvarez [:emilio]]

[Tracking Requested - why for this release]:

My patch in bug 2021155 will fix this and adds a test, but we should make sure we don't ship this if that doesn't make it on time.