110410 - memory leak gone up by 10X (644B to 6880B)

Status: RESOLVED FIXED

(Core Graveyard :: Tracking, defect)

Description

[Cathleen]

Comment 1

[Chase Tingley]

According to tinderbox, this happened between 17:20 and 18:20.

http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=all&branch=HEAD&branchtype=match&dir=&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=11%2F15%2F01+17%3A20%3A00&maxdate=11%2F15%2F01+18%3A20%3A00&cvsroot=%2Fcvsroot

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

This was at least mostly danm, and he temporarily disabled his changes.

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: fix for main (nsDocument) leak, the one that was fixed by disabling danm's changes

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

The other leak is because there's no release of the old mNameSpaceManager before
the NS_NewNameSpaceManager calls in nsDocument::Reset and
nsDocument::ResetToURI.  cc:ing jst since I'm not sure the best way to fix this.
 In particular:
 * it should probably be an nsCOMPtr
 * why do we even need to create a new one for Reset, since it contains no
state?  (Why do we even need one per document?)

Comment 5

[Johnny Stenback (:jst)]

Yeah, throwing away the namespace manager in ::Reset() doesn't make sense,
creating one if there isn't one seems more reasonable.

Comment 6

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Anyway ->danm.

Comment 7

[Dan M]

  Just 10x? I'll try to do better next time.
  Thanks David for tracking it down! And fixing it! Your patch is in, as is a fix 
for the mNameSpaceManager leak.