[css-contain-2] Used value of contain

[andruud]

The spec says that content-visibility can "turn on" layout/style/etc containment, but it's not clear what effect (if any) this has on the computed/used value of contain.

I think the most logical thing would be to specify that if content-visibility is such-and-such, then the used value of contain becomes this-and-that, and specify that the resolved value is the used value.

I bring this up because the Blink implementation of content-visibility modifies the computed value of contain, which seems heavy-handed. Introducing computed-value time dependencies between properties comes at a price. We should avoid doing so unless it's crucial.

cc @vmpstr

[emilio]

I agree, fwiw.

[tabatkins]

I agree it shouldn't alter the computed value, and didn't intend for it to. Do we really need to expose the active set of containments in gCS()? (I'm fine if we do need to, I just want to make sure there's a need and it's not just taken for granted.)

[tabatkins]

(My intention, fwiw, is that it had no effect whatsoever on the 'contain' property. An element having certain containments is just a quality it can possess, and the 'contain' property is one way of applying them, not the only way. But if we think it's useful to communicate the current set of containments, then doing so via the resolved value of 'contain' does indeed seem like the most straightforward way to do so.)

[andruud]

As far as I'm concerned, not affecting contain at any value stage is even better. You're right that I just took it for granted that we wanted to expose the active containments. I don't know if it's really needed. @vmpstr: Any thoughts on that?

My intention, fwiw, is that it had no effect whatsoever on the 'contain' property

By default I would have interpreted the spec that way as well (after all it doesn't explicitly mention messing with the contain value). But the Blink impl. made me doubt what the intent was.

[vmpstr]

As long as the effects of containment still apply when the content-visibility spec says such containment is turned on (which is true), I don't really think we need to change contain value.

There may be a chance that developers rely on this as a way to know whether content-visibility: auto element is currently being skipped or not via presence of size containment. However, I don't think they should be, and the spec never intended that.

It seems like this is a "no change needed" for the spec, and simply a bug in Blink. WDYT?

[andruud]

no change needed [...], bug in Blink

Works for me.