[css-pseudo] Should the "first formatted line" propagate into a different BFC?

[kojiishi]

Split the issue from https://github.com/w3c/csswg-drafts/issues/10990#issuecomment-2394760911 to discuss easier.

Blink I think currently has issues with ::first-line propagating into elements which establish independent formatting contexts, when arguably it shouldn't, when I last read the spec here somewhat questionable. E.g. https://www.software.hixie.ch/utilities/js/live-dom-viewer/?saved=13169

@bfgeek @fantasai

[fantasai]

We need to review the implications for ::first-line, ::first-letter, and text-box-trim.

[css-meeting-bot]

The CSS Working Group just discussed [css-pseudo] Should the "first formatted line" propagate into a different BFC?.

The full IRC log of that discussion

<matthieud> fantasai: text-box-trim feature really on concept of first formatted line
<matthieud> fantasai: also first-line and first-letter rely on this
<matthieud> fantasai: to what extend are the allowed to intrude into a new formatting context like ???
<matthieud> fantasai: we need to think about use cases
<matthieud> fantasai: possible easy answer : they dont (current implementation)
<florian> q+
<dholbert> s/???/flex, grid, multicol/
<matthieud> fantasai: in the past, people uses ??? to create a new formatting context
<Rossen16> ack dbaron
<emilio> q+
<matthieud> dbaron: some case makes sense, other are problematic
<matthieud> dbaron: the weirdest is when a new formatting context has multiple stuff at the top
<matthieud> dbaron: like multi column or grid
<matthieud> dbaron: whether that justify to make it work with all new formatting context is another question
<dbaron> s/work/not work/
<emilio> q- emilio
<dbaron> dbaron: But I think it's weird to apply it to the cases where there are multiple thingns at the top or bottom
<Rossen16> ack florian
<matthieud> florian: by not intruding into nested contexts, you mean that the first formated line wouldn't exist in such cases, or that it would be the first line after the nested formatting context?
<matthieud> fantasai: would not exist
<fantasai> dbaron, I think it might make sense to apply it to all of the things at the top
<dbaron> (This seems different from first-line and fisrt-letter in this regard.)
<fantasai> dbaron, but definitely not just one of them

[Loirooriol]

• to what extend are the allowed to intrude into a new formatting context like flex, grid, multicol

  From https://drafts.csswg.org/css2/#first-line-pseudo

  The :first-line pseudo-element can only be attached to a block container element.

  Back then this only excluded tables, but I don't see why we should treat flex and grid containers different than tables.

  A multicol container establishes an independent formatting context, so it's automatically excluded if these are excluded. I'm fine with excluding multicol containers even if we don't exclude independent FC in general.

• current implementation

  As mentioned in the 1st comment, Blink propagates it to block containers with independent FC, including overflow: scroll, display: flow-root, column-count: 2. For multicol, it only propagates to the 1st column.

  Webkit doesn't propagate to them (but propagates to block containers that continue the BFC of the parent).

  Gecko never propagates to anything.

[kojiishi]

Here is the logic Blink excludes tables, flex, grid, <button>, and button input for ::first-letter.