[css-mixins-1] Are @mixin rules valid within other at-rules?

[andruud]

For example, are @mixin rules valid within @container?

@container (...) {
  @mixin --m() {
    /* style rules */
  }
}

If yes, we already know that --m() isn't defined conditionally based on CQ evaluation (https://drafts.csswg.org/css-conditional-5/#container-rule, last paragraph of chapter), but are the inner style rules affected?

div {
  @apply --m();
}

In other words, does the above @apply rule "remember" that the referenced mixin is enclosed by an @container? Any decision we make should hopefully answer the same question about @layer, @starting-style, etc.

My initial thought is that (for simplicity) we should make @mixin invalid within any at-rule except conditional rules with the "old school" processing model (i.e. currently @media and @supports).

[romainmenke]

My initial thought is that (for simplicity) we should make @mixin invalid within any at-rule except conditional rules with the "old school" processing model (i.e. currently @media and @supports).

I somewhat agree with this.

But I think mixin rules should also be valid in @layer rules to be able to use cascade layers to resolve name conflicts.

However the inner styles should not be affected by this I think.
I would expect :root to have color green in this example:

@layer early-layer, later-layer;

@layer later-layer {
  @mixin --m() { /* later layer declaration wins */
    color: green;
  }
}

@layer early-layer {
  @mixin --m() {
    color: red;
  }
}

:root {
  color: pink;
}

:root {
  @apply --m(); /* inner styles are unlayered */
}

Authors can also layer inner styles by writing:

@layer a {
  @mixin --m() {
    @layer b {
      color: green;
    }
  }
}

[andruud]

But I think mixin rules should also be valid in @layer rules to be able to use cascade layers to resolve name conflicts.

Ah, yeah, you're right. Damn. I was hoping we could just cleanly state "doesn't work (at all) within such and such", but I guess we do need to be more subtle about it---at least for @layer.

However the inner styles should not be affected by this I think.

Hopefully it will still be somewhat explainable to authors that layers affect which mixin definitions "wins", but not the definitions themselves.

[romainmenke]

This also raises some questions about the declared layer order.
Do @layer rules inside mixins affect the layer order?
Or only when those mixins are applied?