[css-display-3] parent box vs anon boxes

[prjnt]

https://drafts.csswg.org/css-display-3/#css-parent-box says:

The parent box of an element’s principal box is (except in the case of run-in) the principal box of its nearest ancestor element that generates a box.

The above implies that an anonymous box is not the parent box of any element's principal box, whereas I believe that it is desired that anonymous boxes can have an element's principal box as a child box, and that the anonymous box should sometimes furthermore establish the containing block for an element's principal box.

[Loirooriol]

I believe that it is desired that anonymous boxes can have an element's principal box as a child box

IMO the problem is that this "parent box" concept has a misleading name. It doesn't refer to the parent node in the box tree. So I agree that anonymous boxes can have children in the box tree, but they are just not taken into account by this "parent box" concept. I don't know whether this is appropriate because I haven't seen any use of the concept.

anonymous box should sometimes furthermore establish the containing block

I'm not that sure. In CSS2, it's true that anonymous boxes can generate a containing block, but then they are skipped when doing percentage resolution:

Anonymous block boxes are ignored when resolving percentage values that would refer to it: the closest non-anonymous ancestor box is used instead.

However, I haven't seen this statement ratified by any CSS3 spec. This makes me think that maybe anonymous boxes can no longer generate a containing block. In fact, before the big rewrite, CSS Box defined containing block like this:

csswg-drafts/css-box-3/Overview.bs

Lines 564 to 568 in 24ef9b6

	<li>Otherwise, if the parent has a <a>principal
	box</a> that is a <a>block container box</a>
	(e.g., if the parent is a 'table-cell' or a 'block'),
	then the containing block is the content edge of the
	parent's <a>principal box.</a>

[prjnt]

Renaming this concept to something other than "parent box" might well be an appropriate fix, assuming that there is indeed prose that intends to refer to this concept rather than parent node in box tree and that such prose can be updated to use the new name.

[Regarding anonymous boxes and percentages, note that the quoted CSS2 visuren.html text is limited to anonymous block boxes: other anonymous boxes, such as anonymous table cells, still establish the containing block in the sense of being used as the basis for resolving percentage widths.]

[fantasai]

No, it's just an error. The definition should match @prjnt's expectations in the OP. :)

[fantasai]

OK @prjnt , I posted some edits. Lmk if it looks correct now or if you have any suggestions for improvement!

[fantasai]

Based on feedback from @Loirooriol in issue 3158, the text now reads

In constructing the box tree, boxes generated by an element are descendants of the principal box of any ancestor elements. In the general case, the direct parent box of an element’s principal box is the principal box of its nearest ancestor element that generates a box; however, there are some exceptions, such as for run-in boxes, display types that generate multiple container boxes (such as tables), and intervening anonymous boxes.