nsFont objects tend to be used on the stack, e.g. by TextFrame in its Paint() member function, and because nsFont contains a nsString object that allocates space from the heap, we end up with a large number of temporary object generated. We either need to not have nsFont allocate the string from the heap, or re-think the usage pattern: e.g., TextFrame is defining a nsFont instance so it can override the decoration that's in the style information

Setting all current Open/Normal to M4.

per leger, assigning QA contacts to all open bugs without QA contacts according to list at http://bugzilla.mozilla.org/describecomponents.cgi?product=Browser

Reassigning qa contact to janc.

Pushing off non-beta 1 issues

Reassigning peterl's bugs to myself.

Accepting peterl's bugs that have a Target Milestone

Triaging my bugs... This one shouldn't be in my list. I had a quick look and below are all the instances of nsFont I found on the stack: nsFormControlHelper::GetFrameFontFM nsFormControlHelper::PaintRectangularButton nsGfxTextControlFrame::Reflow (DEBUG) nsGfxTextControlFrame::InitializeTextControl nsNativeButtonControlFrame::PostCreateWidget nsNativeSelectControlFrame::PostCreateWidget nsNativeSelectControlFrame::PaintSelectControl nsNativeTextControlFrame::PostCreateWidget nsNativeTextControlFrame::PaintTextControl Reassigning to the XPToolkit team.

I don't see why you're reassigning to XPToolkit, none of these are ours. Ignoring the vestigial native widgets, 3 of the remaining 4 occurrences seem to be checked in by rods, reassigning.

Troy, if we made the nsString in the nsFont object be an nsAutoString, does this solve the problem?

No, I don't think we can do that. We looked at that once before and I think Peter embeds nsFont in his style data structs. Using an nsAutoString would cause the size of the style data structs to go way up

moving to M15

reassigning to Kevin

Reassigning back to Rod.

I changed the usage pattern, there are no longer any nsFont object declared on the stack and textFrame does not declare any on the stack either. This will help memory fragmentation

Verified fixt per Troy (Thanks Troy!)

January 14, 2008 ---- Test entry