32448 - [FIX]Viewer leaking widgets

Status: RESOLVED FIXED

(Core Graveyard :: Viewer App, defect, P3)

Description

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

On Linux, viewer is leaking 5 nsBaseWidget every run.  This is a somewhat recent
regression (between 02-21 and 03-18).  The following patch fixes it for me, but
I'm not sure if it's the right thing to do, and I'm not sure if it's OK on all
platforms:

Index: nsBrowserWindow.cpp
===================================================================
RCS file: /cvsroot/mozilla/webshell/tests/viewer/nsBrowserWindow.cpp,v
retrieving revision 3.311
diff -u -r3.311 nsBrowserWindow.cpp
--- nsBrowserWindow.cpp 2000/03/15 01:19:42     3.311
+++ nsBrowserWindow.cpp 2000/03/19 18:19:09
@@ -240,13 +240,13 @@
       }
 
    DestroyWidget(mBack);         
-   mBack = nsnull;
+   NS_IF_RELEASE(mBack);
    DestroyWidget(mForward);      
-   mForward = nsnull;
+   NS_IF_RELEASE(mForward);
    DestroyWidget(mLocation);     
-   mLocation = nsnull;
+   NS_IF_RELEASE(mLocation);
    DestroyWidget(mStatus);       
-   mStatus = nsnull;
+   NS_IF_RELEASE(mStatus);
 
    if(mThrobber)
       {
@@ -256,7 +256,7 @@
       }
 
    DestroyWidget(mWindow);       
-   mWindow = nsnull;
+   NS_IF_RELEASE(mWindow);
 
    return NS_OK;
 }

Comment 1

[rods (gone)]

I'll take this one

Comment 2

[rods (gone)]

I am almost wondering if the release shouldn't go into the DestroyWidget method. 
 But we have to look and see why the mWindow isn't being released.

Comment 3

[rods (gone)]

This fixes it on WinNT and Linux, I have also added asserts to notify us of 
leaks in the future.

Comment 4

[rods (gone)]

changing to M15

Comment 5

[rods (gone)]

fixed

Comment 6

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

Reopening.  I don't think NS_ASSERTION evaluates its arguments in an optimized
build.  Therefore this will still leak in optimized builds, but not debug.  (I
really don't see the need for the assertions.  If things leak, they can be
caught by setting XPCOM_MEM_LEAK_LOG.)

Comment 7

[rods (gone)]

Very good catch, I'll check in the fix tmorrow.

Comment 8

[rods (gone)]

fixed again