67454 - Fix for bug 61756 caused leaks to increase

Status: VERIFIED WONTFIX

(Core :: Graphics: ImageLib, defect)

Description

[Daniel Bratell]

The fix for bug 61756 caused a (or several) memory leaks, which was directly
visble on tinderbox. 


The fix was a one liner that returned early which seemed to have jumped some
cleaning up. I guess this is a dupe, but I couldn't find another bug.

Comment 1

[Daniel Bratell]

According to tinderbox, the following started to leak:

ImgDCallbk                                      144          -
ImageNetContextImpl                             288          -
ImageRendererImpl                                12          -
nsVoidArray                                      72          -

Comment 2

[Gagan]

saari is investigating this. 

Comment 3

[Blake Ross]

saari, would be great if we could get this for .8 (within the next few days)

Comment 4

[saari (gone)]

The leak is timing releated, and of course, when I do hit the new one line of
code, I always see things cleaned up properly later, probably because I'm
changing the timing. Anyone know what speed the leak test machine is?

Comment 5

[Stuart Parmenter]

the code causing this bug is going away, and us wasting time trying to fix this 
is only going to delay us replacing this code.  I don't see any reason to 
continue work on this bug.

Comment 6

[saari (gone)]

I thought I had a fix for this but discovered that I really don't understand the
current imglib ownership model and my fix was totoally ineffective. Rather than
spend more time on this I'm focusing on the new code, bug leaving this bug open
until that new code lands.

Comment 7

[saari (gone)]

will be "fixed" by libpr0n. The leak is in old, end of life code.

Comment 8

[saari (gone)]

libpr0n is on, this is now dead

Comment 9

[Terri Preston]

Verified