[css-cascade-7] What happens with duplicate @sheet identifiers?

[KurtCattiSchmidt]

With the following CSS, what should happen?

@sheet sheet1 {
  * { font-family: serif; }
}
@sheet sheet1 {
  * { font-family: sans-serif; }
}

I can think of a few options:

1. Last definition wins - this aligns with how CSS generally works and seems straightforward for developers.
2. Rules with the same identifier are merged, like with duplicate @layer names. @layer is a bit different though, because source order of the name determines layer precedence, which isn't the case for @sheet.

[romainmenke]

I think this depends on if @sheet is more like a condition (e.g. @media, @layer, ...), or more like a name-defining at-rule (e.g. @keyframes, @font-face, ...).

I think the utility of the append or merge behavior is fairly limited and might be surprising to CSS authors when the sheets contain @import statements.

@sheet sheet1 {
  @import "foo.css";

  * { font-family: serif; }
}

@sheet sheet1 {
  @import "bar.css"; /* won't work, it appears after a style rule when merged */

  * { font-family: sans-serif; }
}

While "last one wins" can also be used by CSS authors to override definitions of sheets.

@sheet sheet1 {
  * { font-family: sans-serif; }
}

@media print { /* If nesting in conditions is even allowed? */
  @sheet sheet1 {
    * { font-family: serif; }
  }
}

[kbabbitt]

Agree with @romainmenke. I think @sheet falls into the "name-defining" category and so I would expect "last one wins" behavior.

[css-meeting-bot]

The CSS Working Group just discussed [css-cascade-7] What happens with duplicate @sheet identifiers?, and agreed to the following:

• RESOLVED: WHere @sheet names conflict, the last one delcared wins

The full IRC log of that discussion

<emeyer> kurt: Yet another @layer divergence, which I used as imspiration for @sheets
<emeyer> …In @layer, duplicate names cause rules to be merged, but here this is up for interpretation
<TabAtkins> q+
<emeyer> …Other name-defining rules are, last one wins, and I think that makes sense here
<emeyer> TabAtkins: I agree with the points in the thread
<astearns> ack TabAtkins
<emeyer> …@layer is setting an aspect of specificity, and not an independent construct
<emeyer> …@sheet does create a construct, and thus last shoudl win
<emeyer> astearns: Is that okay with you?
<emeyer> kurt: Yes
<bramus> (I’m undecided on it)
<emeyer> astearns: So @sheet conflicts go with last-defined wins
<emeyer> bramus: I’m just not sure which way this should swing, but Tab makes a good point
<emeyer> astearns: So your concern is more why @layer does this?
<emeyer> bramus: More that if @layer does this, why doesn’t @sheet? Authors might not see the technical different underneath the two things
<emeyer> astearns: How about a resolution to have @sheet use last-definition-wins for now, without prejudice; we can raise again later if the need arises
<emeyer> …Any objections?
<emeyer> (silence)
<kurt> https://github.com//pull/11980
<emeyer> RESOLVED: WHere @sheet names conflict, the last one delcared wins
<emeyer> kurt: Feedback welcome on #11980 as well
<kurt> Formatted version here: https://kurtcattischmidt.github.io/kurtspublishedw3cdrafts/AtSheet.html

[justinfagnani]

Sheet names should not form a global namespace. That's makes individual CSS files potentially interfere with each other. This would break the bundling use case and isolation between CSS modules.

Can you clarify if the last-rule wins rule applies across files?

[romainmenke]

How would this break the bundling use case?
Can you provide a specific example?

That's makes individual CSS files potentially interfere with each other.

What you are describing is just how CSS works, right?

For my purposes (as maintainer of multiple CSS bundlers) it is actually important that this works exactly as any other CSS feature and that sheet names are shared across multiple files.

Multiple CSS files bundled with @sheet should work exactly the same as when importing each file individually. @sheet should not be an opt-in to some other way of applying multiple stylesheets to a document in my opinion.

Not saying that there isn't a use case for the behavior you are describing, but I am unsure if there is a strong enough case to go against so much precedent of how things work in CSS :)

[justinfagnani]

I guess I'm unclear on the status of the previously proposed global namespace of sheet names, where a sheet could be imported by name regardless of which outer stylesheet it came from.

What I think is important to preserve is that two separate stylesheets nested sheet names are not merged, ie:

Given:

a.css:

@sheet foo {
  p { color: red; }
}

b.css:

@sheet foo {
  p { font-weight: bold; }
}

This applies styles from both foo @sheet rules:

@import foo from "a.css";
@import foo from "b.css";

(ie, bold and red, not just bold);

And these two JS modules get their respective foo sheets independently:

/* a.js */
import {foo} from './a.css';

/* b.js */
import {foo} from './b.css';

A previous version of this proposal at least put sheet names into a global list and there was an open question of how to merge sheets within that list. Both merging and list-wins in that context would cause severe problems.