When viewing the attached testcase, you get this assertion: ###!!! ASSERTION: no last inline frame: 'PR_FALSE', file h:/mozilla/tree-main/mo zilla/layout/base/nsCSSFrameConstructor.cpp, line 9142 Stacktrace: nsDebugImpl::Assertion(nsDebugImpl * const 0x00268170, const char * 0x02113ae8, const char * 0x02113adc, const char * 0x02113a98, int 9142) line 266 NSGlue_Assertion(const char * 0x02113ae8, const char * 0x02113adc, const char * 0x02113a98, int 9142) line 107 nsCSSFrameConstructor::NeedSpecialFrameReframe(nsIContent * 0x03adbbc8, nsIContent * 0x03adbbc8, nsIFrame * & 0x04536eb4, nsIContent * 0x03adbe68, int 2, nsIFrame * & 0x04536fcc, nsIFrame * 0x00000000) line 9142 + 35 bytes nsCSSFrameConstructor::ContentInserted(nsIContent * 0x03adbbc8, nsIContent * 0x03adbe68, int 2, nsILayoutHistoryState * 0x036f6090, int 0) line 9391 + 51 bytes nsCSSFrameConstructor::RecreateFramesForContent(nsIContent * 0x03adbe68) line 11645 + 39 bytes nsCSSFrameConstructor::RestyleElement(nsIContent * 0x03adbe68, nsIFrame * 0x045370a0, nsChangeHint nsChangeHint_ReconstructFrame) line 10527 nsCSSFrameConstructor::ProcessOneRestyle(nsIContent * 0x03adbe68, nsReStyleHint eReStyle_Self, nsChangeHint nsChangeHint_ReconstructFrame) line 13351 nsCSSFrameConstructor::ProcessPendingRestyles() line 13400 nsCSSFrameConstructor::RestyleEvent::HandleEvent() line 13460 HandleRestyleEvent(PLEvent * 0x035a9848) line 13469 PL_HandleEvent(PLEvent * 0x035a9848) line 688 + 10 bytes PL_ProcessPendingEvents(PLEventQueue * 0x00ec2578) line 623 + 9 bytes _md_EventReceiverProc(HWND__ * 0x002501e8, unsigned int 49401, unsigned int 0, long 15476088) line 1408 + 9 bytes USER32! 77e3158f() USER32! 77e31dc9() USER32! 77e31e7e() nsAppStartup::Run(nsAppStartup * const 0x02350d90) line 208 main1(int 1, char * * 0x00262638, nsISupports * 0x00ea1888) line 1248 + 32 bytes main(int 1, char * * 0x00262638) line 1737 + 37 bytes mainCRTStartup() line 338 + 17 bytes KERNEL32! 77e98989() BTW: When you remove a linebreak here or there in the testcase, the assertion disappears. You need to do Shift+Reload to test it again so it does not ask the cache (with cache you do not get the assertion). This bug here might be a regression, but i only have one other debug build to compare to (two month old) and there i do not see this assertion.

I'd hitting this assertion too. I'm not sure the testcase is the same; just guessing that it is.

Still happens on trunk with this testcase (on initial load / shift+reload only).

Probably has to do with the "XXXbz see comments in ContentInserted about this being wrong" comment in this function? In any case, this is a fail-safe situation; I suspect we could just remove the assert... Would still be nice to make the code a little better, though.

Comment on attachment 274356 [details] [diff] [review] This assert is just bogus. Removes bogus assert so that Jesse can test in a bit more peace.

Comment on attachment 274356 [details] [diff] [review] This assert is just bogus. a19=dbaron

Fixed.