1447865 - Implement font-variant descriptor for @font-face rules [was: interaction of font-variant descriptor, font-variant-ligatures property, font-variant shorthand property & font-feature-settings low-level property for ligatures is incorrect]

Status: RESOLVED WONTFIX

(Core :: CSS Parsing and Computation, defect, P3)

Description

[chris]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:61.0) Gecko/20100101 Firefox/61.0
Build ID: 20180321102906

Steps to reproduce:

Run https://test.csswg.org/harness/test/css-fonts-3_dev/single/font-variant-05/format/html5/


Actual results:

Top three lines have crosses instead of check marks for the last two positions. This test is using a special font to check which OpenType features are enabled or disabled.


Expected results:

All lines should look identical.

Setting ‘font-variant-ligatures’ properties will override the same properties set in the
@font-face rule. But properties set there and not explicitly turned off or on later remain set.

Tests interaction of font-variant descriptor, font-variant-ligatures property, font-variant shorthand property & font-feature-settings low-level property for ligatures, in accordance with  steps 1,2,4 and 6 of section 7.2 Feature precedence
https://www.w3.org/TR/css-fonts-3/#feature-precedence

tested in Beta 60 and Nightly 61 on Windows 10

Comment 1

[Ryan Hunt [:rhunt] (on leave until early May)]

Needinfo'ing jfkthame here as I'm not sure what needs to be done here, or what plans are.

Comment 2

[Jonathan Kew [:jfkthame]]

AFAICT, Gecko has never implemented font-variant as a descriptor in @font-face rules (it only exists as a CSS property).

It can hardly "interact correctly" when it doesn't exist! :( Looks like something we just omitted to do, back when font feature support was implemented -- and AFAIK nobody has reported it until now.

Comment 3

[Jonathan Kew [:jfkthame]]

Chris, do you know if any UA actually implements the font-variant descriptor?

Comment 4

[chris]

I have made a PR on web platform tests to check for that
https://github.com/w3c/web-platform-tests/pull/10415

From initial tests, Safari does implement the descriptor, while Chrome and Gecko do not.

If we can get a second implementation this will stay in CSS Fonts 3, otherwise it will get punted to Fonts 4.

Being able to set your preferred default values in a descriptor, which gets applied wherever that font-family is used, is handy. You can override specific ones with the properties, of course.

Comment 6

[Cameron McCormack (:heycam)]

Per https://github.com/w3c/csswg-drafts/issues/2531 this is WONTFIX. Bug 1581729 is for removing the partial support for this (such as FontFace.variant).