368430 - "ASSERTION: unexpected frame type" with text in maction

Status: RESOLVED WORKSFORME

(Core :: MathML, defect)

Description

[Jesse Ruderman]

Attachment: testcase

Loading the testcase triggers:

###!!! ASSERTION: unexpected frame type: 'Not Reached', file /Users/admin/trunk/mozilla/layout/base/nsCSSFrameConstructor.cpp, line 10620

Comment 1

[Boris Zbarsky [:bzbarsky]]

Looks like a bug in nsMathMLmactionFrame::Reflow.  It sets a constrained available height on the single child (which implies that pagination or columnation is happening), then just outputs the child's aStatus.  Then we try to create a continuation for the nsMathMLmactionFrame and that fails.

I _think_ this code wants to use an unconstrained available height...

Comment 2

[Vlad Sukhoy]

Here is some data from nsMathMLmactionFrame::Reflow in the testcase:
<<<<< REFLOW, aReflowState.ComputedWidth = 0
<<<<< REFLOW, aReflowState.ComputedHeight = 1073741824

I don't really understand much what all that means, but zero somehow seems wrong NS_UNCONSTRAINEDSIZE is 1073741824..

Moreover, if I use 
nsSize availSize(NS_UNCONSTRAINEDSIZE, NS_UNCONSTRAINEDSIZE);
in nsMathMLmactionFrame::Reflow()

instead of 
nsSize availSize(aReflowState.ComputedWidth(),
                 aReflowState.ComputedHeight());

the assert's gone, and the rendered page matches the 1.8's one.

Moreover, if I use above in the testcase for bug 378146, the crash is gone..

Comment 3

[Karl Tomlinson (:karlt)]

I'm on getting this assertion (on Linux).
I wonder whether bug 397518 has changed the behavior.

Comment 4

[Karl Tomlinson (:karlt)]

(In reply to comment #3)
> I'm on getting this assertion (on Linux).

That should be: I'm not getting this assertion.

Comment 5

[Jesse Ruderman]

WFM (Mac trunk debug).

Comment 6

[Jesse Ruderman]

Crashtest checked in.