654928 - "ASSERTION: Didn't SaveReflowAndBoundingMetricsFor frame!"

Status: RESOLVED FIXED

(Core :: MathML, defect)

Description

[Jesse Ruderman]

Attachment: testcase

###!!! ASSERTION: Didn't SaveReflowAndBoundingMetricsFor frame!: 'metrics', file layout/mathml/nsMathMLContainerFrame.cpp, line 196

Comment 1

[Jesse Ruderman]

Attachment: stack trace

Comment 2

[Karl Tomlinson (:karlt)]

> (gdb) call nsFrame::DumpFrameTree(0x7fff87dd59b0) 

[...]

>           Frame(maction)(0)@0x7fff87dd5118 {0,0,0,0} [state=0000000000000403] [content=0x7fff87d78680] [sc=0x7fff87d764c8]<
>             Block(maction)(0)@0x7fff87dd59b0 next=0x7fff87dd5280 [content=0x7fff87d78680] {0,0,0,0} [state=0000000000d00402] sc=0x7fff87ab9b90(i=1,b=0) pst=:-moz-mathml-anonymous-block<
>               line 0x7fff87dd5a68: count=1 state=inline,dirty,prevmarginclean,not impacted,not wrapped,before:nobr,after:nobr[0x1] {0,0,0,0} <
>                 Text(0)@0x7fff87dd5210 [run=0x7fff9493b160][0,1,T]  {0,0,0,0} [state=0000000040000402] [content=0x7fff87d78700] sc=0x7fff87dd5e48 pst=:-moz-non-element<
>                   "A"
>                 >
>               >
>             >
>             Frame(mo)(1)@0x7fff87dd5280 {0,0,0,0} [state=0000000000000403] [content=0x7fff87d78780] [sc=0x7fff87dd5910]<
>               Block(mo)(1)@0x7fff87dd57b8 [content=0x7fff87d78780] {0,0,0,0} [state=0000000000d00401] sc=0x7fff87dd5ee8(i=1,b=0) pst=:-moz-mathml-anonymous-block<
>                 line 0x7fff87d6e440: count=1 state=inline,clean,prevmarginclean,not impacted,not wrapped,before:nobr,after:nobr[0x100] {0,0,720,1140} <
>                   Text(0)@0x7fff87dd5480 [run=0x7fff9493b200][0,1,T]  {0,0,720,1140} [state=00000000d0600000] [content=0x7fff87d78800] sc=0x7fff87dd5870 pst=:-moz-non-element<
>                     "B"
>                   >
>                 >
>               >
>             >
>           >

"Frame(mo)" is the selected frame for the maction, but GetPreferredStretchSize
is looking at the first child of the maction, "Block(maction)".

Looks like GetPreferredStretchSize should to handle the
NS_MATHML_EMBELLISH_OPERATOR but
!(NS_MATHML_STRETCH_ALL_CHILDREN_HORIZONTALLY ||
NS_MATHML_STRETCH_ALL_CHILDREN_VERTICALLY)
case more generally than assuming the baseFrame is the first frame.

Comment 3

[Karl Tomlinson (:karlt)]

Attachment: don't assume that the embellished operator is the first child

Comment 4

[Frédéric Wang (:fredw)]

Comment on attachment 531240 [details] [diff] [review]
don't assume that the embellished operator is the first child

Review of attachment 531240 [details] [diff] [review]:
-----------------------------------------------------------------

That looks good.

Apart from maction, there is another case where the embellished op is not the first child: an mrow-like element whose children are one embellished op together with space-like elements. I think the patch fixes the issue for this case too.

fred@debian:~/mozilla/src/layout/mathml$ grep "mPresentationData.baseFrame =" *.cpp
nsMathMLContainerFrame.cpp:      mPresentationData.baseFrame = baseFrame;
nsMathMLContainerFrame.cpp:    mPresentationData.baseFrame = nsnull;
nsMathMLFrame.cpp:  mPresentationData.baseFrame = nsnull;
nsMathMLmactionFrame.cpp:  mPresentationData.baseFrame = mSelectedFrame;
nsMathMLmmultiscriptsFrame.cpp:  mPresentationData.baseFrame = mFrames.FirstChild();
nsMathMLmoverFrame.cpp:  mPresentationData.baseFrame = baseFrame;
nsMathMLmsubFrame.cpp:  mPresentationData.baseFrame = mFrames.FirstChild();
nsMathMLmsubsupFrame.cpp:  mPresentationData.baseFrame = mFrames.FirstChild();
nsMathMLmsupFrame.cpp:  mPresentationData.baseFrame = mFrames.FirstChild();
nsMathMLmunderFrame.cpp:  mPresentationData.baseFrame = baseFrame;
nsMathMLmunderoverFrame.cpp:  mPresentationData.baseFrame = baseFrame;
nsMathMLsemanticsFrame.cpp:  mPresentationData.baseFrame = mFrames.FirstChild();

Comment 5

[Karl Tomlinson (:karlt)]

http://hg.mozilla.org/mozilla-central/rev/7835a2b7db2d
http://hg.mozilla.org/mozilla-central/rev/73a5dcbfeead