Introduce new concepts 'embedded-ltr' and 'embedded-rtl' for the directionality of an element #4338
Conversation
|
No, this is wrong. That's not how bidi works. Embeddedness is not a directionality. |
Embeddedness does change semantics, right? |
|
@fantasai You seemed to suggest embeddedness should be a separate notion out of directionality? Then I guess the spec might mention somewhere else that non-phrasing contents automatically create implicit sembeddedness in bidi semantical sense(like inserting implicit <bdi>). The current spec seems missing it. |
…atting It doesn't make sense to allow bi-di formatting to walk across div.
|
So, here is a much more trivial patch than the previous one. I think all other non-phrasing content elements(<article>, <p>, ...) already imply bdi-like effects by their names, so I only address it for <div>. Does the new patch look okay? @fantasai |
|
I don't think DIV is anything different than P in this respect. Their bidi effects are both already defined in terms of the 'unicode-bidi: isolate' value, so afaict no change is needed. Basically the combination of https://html.spec.whatwg.org/multipage/rendering.html#bidi-rendering (which defines what's supposed to happen) and https://html.spec.whatwg.org/#requirements-relating-to-the-bidirectional-algorithm (which makes it required for all UAs, not just ones that opt into the Rendering chapter) covers what you're asking for already. |
This patch resolves #3905, the specific explanation is in the git commit message.
/acknowledgements.html ( diff )
/grouping-content.html ( diff )