[css-view-transitions-2] Clarifications around view-transition-group

[noamr]

We recently resolved on view-transition-group, but a few things remained unresolved:

• tree scoping: assuming that the nearest keyword and others are tree scoped like names. Verifying. ([css-view-transitions-2] view-transition-group and tree-scoping #10633)
• mismatch: what happens when the old/new view-transition-group are not the same? ([css-view-transitions-2] Nested transitions: what happens if there is a mismatch? #10631).
• Does nearest and <<custom-ident>> implicitly also mean contain? ([css-view-transition-2] Should non-default view-transition-group act like contain? #10780)

Proposing:

• view-transition-group semantics are tree-scoped
• view-transition-group acts like view-transition-class in a mismatch (last capture wins)
• nearest and <<custom-ident>> implicitly also main contain

[fantasai]

For the last one, I think that was already resolved (by adopting the proposal as expressed in the issue prior to the resolution).

[css-meeting-bot]

The CSS Working Group just discussed [css-view-transitions-2] Clarifications around `view-transition-group` .

The full IRC log of that discussion

<TabAtkins> noamr: we resolved in the VT breakout on many things, but there were some unclear or edge cases
<TabAtkins> noamr: small but related
<TabAtkins> noamr: want to clarify the semantics of VT keywords are tree-scoped
<TabAtkins> noamr: We have keywords like 'nearest'
<khush> q+
<TabAtkins> noamr: Want to clarify they're tree-scoped as well, so 'nearest' can't capture things that couldn't be explicitly named
<TabAtkins> noamr: the other one is what happens in a mismatch, the old doc says "VT group X" the new says "VT group Y"
<TabAtkins> noamr: It appears there's no good solution
<TabAtkins> noamr: want to just be consistent with v-t-class and such. the semantic there is "last capture wins", so it'll usually take the VT group from the new doc state, unless it's an exit transition
<TabAtkins> noamr: third thing, not clear when using 'nearest' or a name that it also contains its descendants
<TabAtkins> noamr: if not, it's impossible for an element to contain and be contained at the same time
<TabAtkins> noamr: our proposal is to have it also contains its descendants by default
<TabAtkins> noamr: you can always escape by normal
<khush> q?
<TabAtkins> fantasai: it seems like this is three issues all filed separately, can we take them one by one?

[astearns]

working on the individual issues instead