Repository navigation
[css-variables-1][css-values-5] Define "arbitrary substitution function" #10679
Description
In Variables, we have a lot of text defining how to handle a function that can be replaced by an arbitrary token sequence. In Values 5, we're adding several more functions that act similarly. I ended up just referencing the (undefined) concept of an "arbitrary substitution function", and had an issue saying I needed to define it in Variables at some point. Mixins also needs to reference this concept, for custom functions.
Well, I've defined it (in the commit that'll be referenced from this issue in a moment). Please review?
Notably, the only actually new text is the algorithms for "resolve a var() function" and "substitute an arbitrary substitution function"; the rest of the commit is just rearranging text and rephrasing some things to refer to the new generic term rather than be var()-specific.
Assuming this looks good, I'll need to follow up with some edits to Values 5 and to Mixins, to define resolution algorithms.
(Because of the text movement, reviewing the diff might be difficult; I recommend just checking the live Variables spec.)
Activity
[css-variables-1] Define 'arbitrary substitution function', and rearr…
cdoublev commentedon Aug 3, 2024
UAs (at least Chrome and FF) already accept
var()andenv()in descriptors of page and margin rules, includingsize, ie. a descriptor whose name and grammar do not match those of a property.I suggest that:
<env()>should be valid for all declarations<var()>should be valid for all declarations in element and cascade-dependent contexts<attr()>,<random-item()>, should be valid in element-dependent contexts(edit) Reference for the next comment: #5969 (comment)
andruud commentedon Aug 7, 2024
@tabatkins I read through it, LGTM.
@cdoublev Huh. Looking briefly at the Blink code (StyleResolver::StyleForPage), I think
var()would resolve against the root element. I don't think that's per spec.It's a separate issue, though.
css-meeting-bot commentedon Nov 7, 2024
The CSS Working Group just discussed
[css-variables-1][css-values-5] Define "arbitrary substitution function".The full IRC log of that discussion
<florian> scribenick: florian<fantasai> https://drafts.csswg.org/css-values-5/#arbitrary-substitution
<florian> fantasai: we do a bunch of substitutions, that check at computed value time it is valid
<florian> TabAtkins: that's not appendix A in values 5
<florian> s/not/now
<TabAtkins> whoops
<TabAtkins> i'll be right there
<florian> fantasai: just asking if that's fine
<fantasai> https://github.com//issues/10679#issuecomment-2266364116
<florian> fantasai: other questions can be handle as a follow up question
<florian> fantasai: guillaume asked about where these are valid
<florian> fantasai: we might need to make that context dependent, based on which functions
<florian> fantasai: that would be a follow up issue
<florian> astearns: this is just spec definitions?
<florian> fantasai: yes, for var. makes things more reusable
<florian> fantasai: should not result in any changes in existing specs
<fantasai> s/specs/features/
<TabAtkins> astearns: glad Guillaume has already looked at this, he's a graet reviewer
<TabAtkins> astearns: shall we leave it there and go on?
<TabAtkins> scribe+
kbabbitt commentedon Nov 7, 2024
Definition LGTM, the one suggestion I have would be to add a note clarifying the interaction with
!important, similar to the second note under https://drafts.csswg.org/css-variables-1/#syntaxtabatkins commentedon Nov 7, 2024
Hm, that's specifically a note about custom properties, not about the var() itself. It wouldn't be relevant to any other substitution function.
[css-values-5] Replace the notion of arb-sub functions creating a dep…