690864 - dromaeo CSS regression on 32-bit Linux (suspected due to PRBool --> bool conversion)

Status: RESOLVED WONTFIX

(Core :: General, defect)

Description

[Josh Aas]

Attachment: graph

Comment 1

[Boris Zbarsky [:bzbarsky]]

mwu ran some numbers;  http://dromaeo.com/?id=151464,151456 is the before and after.

Looks like something in extjs is just not happy; mwu is looking into it.

Comment 2

[Michael Wu [:mwu]]

I ran this again on a windows vm:

http://dromaeo.com/?id=151828,151829

MSVC appears to be smarter about this.

Comment 3

[(no longer active)]

Has there been any progress in tracking this down?  The longer that this is not fixed, the harder it will get to backout the PRBool patch.  :(

Comment 4

[Boris Zbarsky [:bzbarsky]]

No.  I think that we've decided that since MSVC is not affected and the problem is sort of impossible to reproduce while profiling we're not going to worry about it for the moment...

Comment 5

[(no longer active)]

(In reply to Boris Zbarsky (:bz) from comment #4)
> No.  I think that we've decided that since MSVC is not affected and the
> problem is sort of impossible to reproduce while profiling we're not going
> to worry about it for the moment...

So that means that this is WONTFIX then?  Should I proceed with my PR_TRUE/PR_FALSE elimination project?

Comment 6

[Boris Zbarsky [:bzbarsky]]

I thinks so, on both counts... Michael?

Comment 7

[Michael Wu [:mwu]]

I agree. My best guess is that msvc is being smarter about int->bool conversions than gcc..