Created attachment 30544 [details] [review]
2.6 settings daemon leaks

Could someone please review and apply this patch. Having leaks in the longest
running gnome daemon is a pretty bad bug.

Please review the patch.

Comment on attachment 30544 [details] [review]
2.6 settings daemon leaks

>-	if (!strcmp (star_app->id, plain_app->id))
>+	}
>+	if (!strcmp (star_app->id, plain_app->id)) {
>+	        if (star_app != NULL) {
>+	            gnome_vfs_mime_application_free (star_app);
>+	        }
>+	        if (plain_app != NULL) {
>+	            gnome_vfs_mime_application_free (plain_app);
>+	        }
> 		return;
>+	}

This should probably just be:

>-	if (!strcmp (star_app->id, plain_app->id))
>+	}
>+	if (!strcmp (star_app->id, plain_app->id)) {
>+		gnome_vfs_mime_application_free (star_app);
>+		gnome_vfs_mime_application_free (plain_app);
> 		return;
>+	}

So, should this patch be applied?

Created attachment 31308 [details] [review]
updated patch integrating Vincent's comments

Vincent's advice integrated in this new version of the patch.

Punting to 2.8.x

Thomas, did you run things under valgrind or something similar after doing these
changes? Just want to make sure we're not introducing regressions in this area.
We should really get this in ASAP though to have it tested as much as possible
before the first point release.

I used valgrind to locate the problems in the first place by running the
gnome-settings-daemon into an Xnest display.

Can we get this into CVS soon?

this patch is pending for a while, permission to commit to head ?

Yah, lets get it in.  The changes look sane.  But there are alot of corner cases
in there.  HEAD only please.

patch commit to HEAD, thanks for your contribution !