501035 - Inconsistent layout with <select style="display: table-caption">

Status: NEW

(Core :: Layout: Tables, defect)

Description

[Jesse Ruderman]

Attachment: testcase (dynamic)

Comment 1

[Jesse Ruderman]

Attachment: reference (static)

Comment 2

[Jesse Ruderman]

Based on layout/base/crashtests/341858-1.html

Comment 3

[Daniel Holbert [:dholbert]]

FWIW: On my machine, a normal reload will make the testcase render correctly (i.e. it'll match the reference).  A full reload (with 'shift' key) will make the testcase go back to its buggy rendering.

Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.2a1pre) Gecko/20090624 Minefield/3.6a1pre

Comment 4

[Jesse Ruderman]

I noticed that too.  My guess is that form history is triggering a reflow on a normal reload.

Comment 5

[Bernd]

smells like nsCSSFrameConstructor::IsValidSibling

Comment 6

[Bernd]

Jesse is the sibling action required or would that also work when modifying the span directly?

Comment 7

[Jesse Ruderman]

I can't get the same thing to happen by modifying the span.

Comment 8

[Boris Zbarsky [:bzbarsky]]

OK, I can reproduce the bug with 2009-06-24 but not with 2009-06-23.

Regression range:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=28aa23105a9e&tochange=5fe89f2c22f0

Probably bug 495385.

Comment 9

[Boris Zbarsky [:bzbarsky]]

Ah, yes.  With that patch, the behavior is the same as if you were inserting the textnode there instead of just changing the text.  You should be able to reproduce the bug by doing that even in builds before roc landed.

Bernd, we really need a better solution for the caption thing....

Comment 10

[Bernd]

> Bernd, we really need a better solution for the caption thing....
yep, that is after BC, which means ETA Q1-Q2 2010