334148 - "ASSERTION: wrong kind of child frame" involving float:right, display:list-item, and <xul:menupopup>

Status: RESOLVED WORKSFORME

(Core :: Layout, defect)

Description

[Jesse Ruderman]

###!!! ASSERTION: wrong kind of child frame: 'aIsBlock == f->GetStyleDisplay()->IsBlockLevel()', file /Users/admin/trunk/mozilla/layout/generic/nsLineBox.cpp, line 74

Comment 1

[Jesse Ruderman]

Attachment: testcase

Comment 2

[Boris Zbarsky [:bzbarsky]]

Basically, the binding allows styling by the document...  So we end up with a nsAutoRepeatBoxFrame that has:

1)  display:list-item
2)  float:right

and is being inserted into the in-flow child list of its parent(!).

Note that after the assert this bug is on we get:

###!!! ASSERTION: How'd we get a floated inline frame? The frame ctor should've dealt with this.: 'aReflowState.mStyleDisplay->mFloats == NS_STYLE_FLOAT_NONE', file ../../../mozilla/layout/generic/nsLineLayout.cpp, line 1225

Comment 3

[Boris Zbarsky [:bzbarsky]]

Note that in this case we hit the following code in the frame constructor:

6499     // xul does not support absolute positioning
6500     nsIFrame* geometricParent = aParentFrame;

That's probably bogus; definitely when aParentFrame is not a XUL frame.  I have no idea what the behavior of floated or positioned XUL should be in general, however...

Comment 4

[[e-mail]]

Well I don't see how float can apply to XUL; as for absolute positioning xul already has a way to achieve that so a second way is unnecessary but unharmful.

Comment 5

[Boris Zbarsky [:bzbarsky]]

> Well I don't see how float can apply to XUL;

  <html:div>
    <xul:hbox style="float:left"/>
  </html:div>

That's basically what happens in the testcase, fwiw.

As in, the problem is not with "all XUL" the problem is when XUL interacts with normal CSS layout.

Comment 6

[[e-mail]]

Sorry, I had assumed the parent was XUL; floating XUL in HTML is fine with me.

Comment 7

[Boris Zbarsky [:bzbarsky]]

So if we had an IsFrameOfType() flag for XUL we could probably test the parent here, and if the parent is non-XUL we'd do the normal floating/positioning thing....  Seem reasoanble?

Comment 8

[Boris Zbarsky [:bzbarsky]]

That wouldn't work so hot if parent is an inline whose parent is XUL, actually...

Sounds to me like XUL should push a null float containing block while processing children.  Possibly also a null absolute containing block.  Thoughts?

Comment 9

[Jesse Ruderman]

WFM on trunk.  I see

###!!! ASSERTION: no common ancestor at all???: 'parent', file /Users/jruderman/trunk/mozilla/layout/base/nsLayoutUtils.cpp, line 356

but that's a very frequent assertion and is covered by other bugs.

Comment 10

[Jesse Ruderman]

Crashtest checked in.