[css-text-decor] Consider adding an `all` value to `text-decoration-skip-ink`

[jfkthame]

With text-decoration-skip-ink: auto, ink-skipping is applied to most content, but not to CJK text. This makes sense for common fonts, where the default underline position, designed to work with Latin glyphs, clashes with so many CJK glyphs that skip-ink behavior largely destroys underlining.

This means, however, that it is impossible for an author to specify that they actually do want skip-ink behavior to be used with CJK text. This might be desired in cases where the underline position (either specified by the font, or explicitly controlled by the author using text-underline-offset) is such that it does not clash with most glyphs, but only with occasional exceptionally-deep shapes. With more calligraphic font styles, in particular, the glyphs may not all sit solidly on the same baseline as tends to be the case in square printed styles.

I propose there should be an all value (or always, or other name to be bikeshedded) for text-decoration-skip-ink. The initial value would remain auto, so that the browser will do “the right thing” by default, but all makes it possible for the author to opt in to ink-skipping behavior even for CJK glyphs where auto would disable it.

See the right-hand example here, illustrating how skip-ink behavior might reasonably be applied even to CJK text when using a calligraphic font:

[upsuper]

I wonder what should happen if the font failed to load and it fallback to some normal font.

This makes me think it should probably be controlled by @font-face or so somehow rather than this property.

[drott]

I support the proposal for a new non default keyword to include CJK in ink-skipping, such as all. I don't think there is a need to make it dependent on the success of the styled primary font-family to be available. As jfkthame@ explained, if CJK in-skipping is desirable, it would have to be explicitly selected by placing the new keyword, and potentially in addition to styling the text-underline-offset. I think these two options give reasonable control to avoid an accidental, undesirable application of ink-skipping to the text decoration.

[litherum]

The presence of text-underline-offset makes this a good idea.

[css-meeting-bot]

The CSS Working Group just discussed [css-text-decor] Consider adding an `all` value to `text-decoration-skip-ink` , and agreed to the following:

• RESOLVED: Add an 'always' value to 'text-decoration-skip-ink'

The full IRC log of that discussion

<dael> Topic: [css-text-decor] Consider adding an `all` value to `text-decoration-skip-ink`
<dael> github: https://github.com//issues/4277
<dael> myles: Right now when you turn on text-decoration-skip-ink you can use auto which is skip everything except cjk. cjk letters usually lower on line and intersect too much with underline. That was a decision I made many years ago; looked at cjk and it was awful so made it not apply
<dael> myles: Other value was none which is don't skip.
<dael> myles: Now that there's a property to move underline it's reasonable for an author to want to place underline such that it doesn't intersect. Then having ink-skipping on cjk makes sense. Let's add a 3rd value that does it on everything, even cjk. Makes total sense to me
<fantasai> I agree with doing this, would suggest using always rather than all
<dael> jensimmons: Makes sense to me to give authors rest of power to do what they want. They should have option to force it on even if default for script is off
<dael> florian: Makes sense to me
<dael> astearns: fantasai prefers always instead of all? Any other opinion?
<dael> astearns: Prop: Add an always value to text-decoration-skip-ink
<dael> astearns: Objections?
<dael> RESOLVED: Add an 'always' value to 'text-decoration-skip-ink'

[jensimmons]

I just commented on the call that I think this is a good idea. Just thought I'd add that this will help with teaching as well. Having none and always makes auto make more sense. It's something in-between. It's clearer that auto means the browser will decide, sometimes yes, sometimes no.

[jensimmons]

In typing my last comment, I wondered about pairing:
none and always

Wouldn't it make more sense for a pair of words to be
none and all
or
never and always

It's a grammatical mismatch that will make it hard for authors to remember the yes do it all the time keyword.

@fantasai Why did you want to use always instead of all?