Second jump in mail leaks today: http://bonsai.mozilla.org/cvsquery.cgi?module=MozillaTinderboxAll&date=explicit&mindate=1014096720&maxdate=1014099959 --NEW-LEAKS-----------------------------------leaks------leaks%----------------------- GlobalWindowImpl 200 - nsCodebasePrincipal 60 - nsXPCComponents_Interfaces 20 - BackstagePass 16 - nsJSRuntimeServiceImpl 20 - nsGenericFactory 16 - nsJSID 32 - AtomImpl 24 - nsSimpleURI 4 - nsXPCComponents_ID 12 - nsLocalFile 432 - nsXPCComponents_Results 12 - si_SignonUserStruct 8 - nsJSIID 100 - XPCWrappedNativeProto 224 - xptiWorkingSet 52 - si_SignonURLStruct 16 - nsSupportsArray 52 - nsSystemPrincipal 48 - nsPrefBranch 64 - nsHashtable 88 - xptiInterfaceInfo 80 - nsXBLDocGlobalObject 24 - si_SignonDataStruct 584 - nsAggregatePrincipal 68 - xptiInterfaceInfoManager 152 - NavigatorImpl 24 - nsDOMClassInfo 112 - XPCNativeScriptableInfo 64 - nsComponentManagerImpl 160 - nsXPCComponents_Classes 12 - nsPrefService 36 - nsTimerImpl 36 - SharedScriptableHelperForJSIID 8 - xptiFile 3668 - morkObject 40 - nsEntropyCollector 1044 - XPCWrappedNative 572 1200.00% XPCNativeScriptableShared 836 1000.00% nsStr 204 466.67% nsVoidArray 88 266.67% nsXPCComponents 132 200.00% nsThread 48 100.00%

I'm pretty sure it's not mine. My back end changes were backed out due to orangeness on btek and comet and the front end error reporting changes will never be executed because, right now, the back end will never return an error.

It wasn't me. My checkin was a string change to a feature that's turned off. (time bomb)

The first jump in mail leaks I think was due to jband's checkin. There already is a long discussion about that. Apparently it is from an existing leak which is now holding on to more objects down in xp-connect. That's the jump from 5kb to 9kb. As Chris mentioned when he filed this bug, this bug is tracking the 9kb to 14kb jump. Just stating the obvious.

From what I can tell, the only code checked in during this window that has a chance to effect leaks for the mail tests would be: ducarroz's mime checkins and gordon's cache checkins

Are we talking about the same leaks as bug 126455, or are these leak numbers from a different test?

Jean-Francois backed out his code and it had no effect on the leak numbers. Thanks for trying JF. Feel free to check your mime changes back in. That leaves us with the cache changes. Ahh I just read gordon's comments. Yes indeed they look like the same set of leaks that are already reported in bug 126455. Note the checkin logs for this leak and the one on brad are nearly identical. I'm sure this is a dup. *** This bug has been marked as a duplicate of 126455 ***

verified (it was gordon, apparently).

re-opening to re-resolve this.

dup of 102229, by way of dup bug 126455 *** This bug has been marked as a duplicate of 102229 ***