85872 - Inheritance from elements with display:table-cell broken

Status: VERIFIED FIXED

(Core :: CSS Parsing and Computation, defect)

Description

[Hixie (not reading bugmail)]

The 'content' property doesn't inherit from a cell to it's ':after' pseudo-
element. This works fine for other elements.

Testcase: http://www.hixie.ch/tests/adhoc/css/inheritance/content/002.xml

Comment 1

[Emlyn Shannon]

You should have included the build id of the version you saw this on.

This bug does occur on build 2001060712 on Macintosh.

Comment 2

[Greg K.]

Setting All/All per comment #1.

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Are there other inheritance problems in the same place?

Comment 4

[Boris Zbarsky [:bzbarsky]]

Attachment: Testcase showing border inheritance problem too

Comment 5

[Boris Zbarsky [:bzbarsky]]

So... The problem is that table cells have this block frame inside... and this
block frame is what's used as the parent frame when ProcessChildren() is called
from ConstructTableCellFrame().  So the pseudo style context of that block is
what's used as the parent style context for the <td>'s kids.

Comment 6

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Yet another bug resulting from the backwards relationship between style contexts
and frames.

Comment 7

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I think Boris just fixed this for the dynamic case in bug 289517 but not for the static case.

Comment 8

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Actually, I backed that part of the patch out.  (And I'm actually not 100% sure it's right, although the difference is irrelevant unless somebody styles ::-moz-cell-content.)

Comment 9

[Bernd]

I filed bug 323656 as table cells are not the only case when this happens

Comment 10

[Boris Zbarsky [:bzbarsky]]

Fixed by checkin for bug 323656.

Comment 11

[Jeff Walden [:Waldo]]

I don't see any :after tests in attachment 261617 [details] [diff] [review], so this testcase isn't quite redundant with tests in there and thus should be committed, correct?

Comment 12

[Boris Zbarsky [:bzbarsky]]

I suppose.  I didn't want to commit these tests as-is, because they would break if we implement the proposed CSS3 generated content stuff.  I suppose we could add explicit tests that :before and :after inherit from their parent; I should write some like that.

Comment 13

[Boris Zbarsky [:bzbarsky]]

Though that's going to take weeks, at least; if someone gets to it before me I'll be very very happy.

Comment 14

[Martijn Wargers (dead)]

Verified fixed on trunk and branches.
Both testcase (url and attached one) now show "PASSED" at the places where it is expected.