374038 - [FIX]Canvas layout in XUL broken with XulRunner 1.9

Status: VERIFIED FIXED

(Core :: Graphics: Canvas2D, defect, P2)

Description

[Michael Bonfils]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1.2) Gecko/20070219 Firefox/2.0.0.2
Build Identifier: http://ftp.mozilla.org/pub/mozilla.org/xulrunner/nightly/2007-03-14-08-trunk/

With all releases of XulRunner 1.9 (a1 throught a3), the testcase show a little window with no content instead of a little text with a canvas.

Reproducible: Always

Steps to Reproduce:
1. Open the testcase

Actual Results:  
The content of windows show the background content and the canvas has a size of zero. http://murlock.org/xul/xul_19a3.png

Expected Results:  
With XulRunner 1.8.0.4, the windows look like that : http://murlock.org/xul/xul_1804.png

It may be related to http://bugzilla.mozilla.org/show_bug.cgi?id=366616

Comment 1

[Michael Bonfils]

Attachment: Show the bug

Comment 2

[Dave Townsend [:mossop]]

This looks to be unrelated to the canvas element to me. In the screenshot given no part of the window displays, not even the description at the top.

How are you opening this window?

Comment 3

[Martijn Wargers (dead)]

This is definitely a bug. With the landing of the reflow branch, the canvas element doesn't show up anymore.
Between 2007-01-16 and 2007-01-17, the whole page doesn't get shown at all, instead the previous contents stays painted. This is rather bad.

Comment 4

[Martijn Wargers (dead)]

Attachment: testcase2

Comment 5

[Boris Zbarsky [:bzbarsky]]

Attachment: Possible approach

There are basically two parts to this patch:

1)  Make BoxReflow() actually compute an mComputedHeight using ComputeSize as
    needed.  This is necessary because replaced elements just use the
    mComputedHeight as their height, and the reflow state is not calling
    ComputeSize because it thinks it's a reflow root (no parent reflow state).
    I suppose this call could also be moved into the reflow state.
2)  Change the computation of the boxmetrics mBlockMinSize to set _something_
    nonzero for non-block kids of XUL boxes.  Using whatever height came back
    from BoxReflow in this case seems to make sense to me.  If I don't do this,
    then I get a zero-height canvas, because after reflowing the kid sprocket
    layout clobbers its rect with the stuff hanging out in the boxmetrics.

David, thoughts?  It's a little hacky, but....

Comment 6

[David Baron :dbaron:]

Comment on attachment 260210 [details] [diff] [review]
Possible approach

r+sr=dbaron

Comment 7

[Boris Zbarsky [:bzbarsky]]

Attachment: Regression test

Comment 8

[Boris Zbarsky [:bzbarsky]]

Fixed.

Comment 10

[Samuel Sidler (old account; do not CC)]

v.fixed

Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9pre) Gecko/2008050704 Minefield/3.0pre