[css-writing-modes-4] Allow sideways-x and vertical-x to share a block formatting context?

[fantasai]

Writing Modes currently says that if a block has a different writing-mode value than its parent, then it forms a new block formatting context.

This was quite appropriate for Writing Modes L3, but in L4 we introduced the sideways-* variants that are just like their vertical-* variants except the line orientation is different. This probably doesn't need a new block formatting context.

Should we loosen this restriction so that sideways-* and vertical-* blocks can share a block formatting context?

CC @kojiishi @MurakamiShinyu

[Loirooriol]

CC @tkent-google

[jfkthame]

Should we loosen this restriction so that sideways-* and vertical-* blocks can share a block formatting context?

I think so; there's no reason for the orientation to require a new BFC.

[dbaron]

Not necessarily a blocking issue, but it's worth noting that according to the mappings, the inline-start and inline-end mappings swap between vertical-lr and sideways-lr. This seems like it might be at least a little awkward for the logical values of float and clear and would at least require that the definitions be carefully written to handle the case that those mappings can change within a single BFC.

[fantasai]

I think those definitions are already clear: they're mapped based on the containing block. It's the same problem as if you swapped the direction property.

[dbaron]

I think this seems like a reasonable idea assuming it isn't a compatibility problem. I'd note that according to Chromestatus the sideways-* keywords are scheduled to ship in a Chrome release that has already branched, so it's not clear if Chrome would be able to do this in the first release with these keywords. (Firefox already ships them.) cc: @tkent-google

[tkent-google]

Do web authors want BFC sharing in such case? If not, I don't think we need to change the specification.

[css-meeting-bot]

The CSS Working Group just discussed [css-writing-modes-4] Allow sideways-x and vertical-x to share a block formatting context?, and agreed to the following:

• RESOLVED: Allow consistent sideways-* and vertical-* writing modes to share a BFC

The full IRC log of that discussion

<TabAtkins> fantasai: WM3 said if you have a different writing mode, you form a different bfc
<TabAtkins> fantasai: made sense then, each wm went in a different block-flow direction
<TabAtkins> fantasai: but in WM4 we introduced the sideways variants, which differ from vertical only in line orientation
<TabAtkins> fantasai: so switching between thos edoesn't change block flow direciton, probably doesn't need a new bfc
<TabAtkins> fantasai: so does it make sense to allow them to share a bfc?
<TabAtkins> fantasai: in terms of use-cases, say an english block quote inside of vertical japanese
<TabAtkins> fantasai: and you formatted that with sideways-*
<TabAtkins> astearns: so firefox ships the sideways keywords. does it establish a new bfc there?
<TabAtkins> fantasai: haven't checked, but also don't think we're under compat pressure here
<TabAtkins> jfkthame: i think it does establish a bfc, but think we could change that without too much difficulty
<TabAtkins> astearns: and ???'s comment is that we shoudln't change the spec unless there's a good author use-case
<astearns> s/???/Kent/
<TabAtkins> fantasai: i think if you do mix these two... the cases where there is a behavior difference are rare
<TabAtkins> fantasai: but if you did end up in such a sitaution, you'd want them to be in the same bfc
<fantasai> TabAtkins: obvious example is a float. If you started in Japanese, and then switched to sideways English, and float intruded into the English paragraph
<TabAtkins> astearns: so proposal is to change the spec to not require a new bfc between sideways-* and vertical-* WM changes
<TabAtkins> astearns: and we'll see if it's web compatible
<oriol> The author can always force a bfc manually if desired
<TabAtkins> fantasai: yes. i suspect it is since firefox is the only one shipping sideways rn
<TabAtkins> astearns: comments or objections?
<TabAtkins> RESOLVED: Allow consistent sideways-* and vertical-* writing modes to share a BFC

[dholbert]

Two testcases, maybe (though I might be getting this wrong, since I don't often think about the subtleties of sideways):

data:text/html,<!DOCTYPE html><div style="writing-mode: sideways-rl">AAA<span style="writing-mode: vertical-rl">BBB<span style="float:left">Float

data:text/html,<!DOCTYPE html><div style="writing-mode: sideways-lr">AAA<span style="writing-mode: vertical-lr">BBB<span style="float:left">Float