[css-contain] container shorthand syntax is confusing.

[emilio]

https://andreubotella.com/csswg-auto-build/css-contain-3/#container-shorthand says:

<'container-name'>? [ / <'container-type'> ]?

But tests expect <container-type> [ / <container-name> ]? which makes a lot more sense to me. Having both parts of the syntax optional doesn't really make sense to me.

cc @mirisuzanne @andruud

[SebastianZ]

And if both parts of the syntax should be optional, the syntax is still incorrect.

If all parts of a group are optional but need to produce at least one value, you need to add an exclamation mark, but in this case we have the slash before the second value. That would make values like / style valid.

So it should be:

<'container-name'> [ / <'container-type'> ]? | <'container-type'>

Having said that, this syntax would be ambiguous as names could clash with the types. So again, if container names are also be meant to be optional, the prose needs to clarify that the keywords defined for container-type are invalid as container names.

Sebastian

[andruud]

Duplicate? #7142

[mirisuzanne]

Yes, sorry, the spec (and tests) are out-of-date with the current resolutions (because I thought we were revisiting a decision, but then we didn't). I'll get those changes in the spec today/tomorrow. Sorry for the confusion.

In brief:

• We resolved that the initial value of container-type is style, not none.
• In the same resolution, since both name & type are now 'optional' for a container, we swapped the order of the container shorthand to encourage naming containers.
• (we also swapped the order in @container so they would match, but then we removed container-type from that syntax altogether)

So, in relation to the shorthand syntax that means currently:

• If no container-name is provided, it is set to none ([css-contain-3] container :/ size #7142 suggests making this explicitly required)
• If no container-type is provided, it is set to style

[mirisuzanne]

The Editor's Draft is now updated to reflect the current state of our resolutions on this. There are open issues for addressing the fact that both our optional (linked above), and your request to revisit the style initial-value. Closing as question answered for now.