Jesse Ruderman Reporter
	Description • 17 years ago

Loading the testcase triggers:

###!!! ASSERTION: assigning special sibling to other than first-in-flow!: '!aFrame->GetPrevInFlow()', file /Users/jruderman/trunk/mozilla/layout/base/nsCSSFrameConstructor.cpp, line 490

The final layout is consistent with the same DOM loaded as markup, but Find for the inserted text is broken when the bug is triggered.  I took advantage of that by calling find() in order to make a reftest pair.

	Jesse Ruderman Reporter
	Comment 1 • 17 years ago

In the reference, "XXX" is selected correctly, but in the testcase, it isn't (due to the bug that causes the assertion to fire, I'm guessing).

	Boris Zbarsky [:bzbarsky] Assignee
	Comment 2 • 17 years ago

Jesse, that is a _nice_ test!

	Boris Zbarsky [:bzbarsky] Assignee
	Comment 3 • 17 years ago

Comment on attachment 278165 [details] [diff] [review]
Fix

Very safe regression fix.

	David Baron :dbaron:
	Comment 4 • 17 years ago

Comment on attachment 278165 [details] [diff] [review]
Fix

a1.9=dbaron

	Boris Zbarsky [:bzbarsky] Assignee
	Comment 5 • 17 years ago

Fixed.  Test checked in.

	Jesse Ruderman Reporter
	Comment 6 • 17 years ago

The tests were checked in as .html files.  I think they should have been checked in as .xhtml, since they use div inside span (the HTML parser doesn't like that).

	Boris Zbarsky [:bzbarsky] Assignee
	Comment 7 • 17 years ago

Uh... no.  The HTML parser allows <div> inside <span> just fine.  I tested with the testcases I checked in before and after the patch, of course.

	Jesse Ruderman Reporter
	Comment 8 • 17 years ago

Oh, interesting.  It does strange things with <div> inside <b>, but it leaves <div> inside <span> alone.

	Boris Zbarsky [:bzbarsky] Assignee
	Comment 9 • 17 years ago

Yep.