I'm trying to figure out why browser_bug435788.js leaks pluginInstallerWizard.xul 10 times. The longwindedness of the test doesn't help, so I'd like to consolidate this. When I pushed the attached patch to the tryserver, the number of leaked wizard windows went down 9, but it could be that that number just isn't stable.

> When I pushed the attached patch to the tryserver, the number of leaked > wizard windows went down 9 It went down /to/ 9 (not "by", which would have been great).

I looked through a few more logs, the number of leaked windows is more or less random (varied between 7 and 12).

got rid of eval

Comment on attachment 539158 [details] [diff] [review] patch v2 Review of attachment 539158 [details] [diff] [review]: ----------------------------------------------------------------- Looks ok. I heard rumours that we are ditching the pfs code, if that is true then spending much time on this might be a waste of time. Dolske might know more.

Verified fixed based on results in http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1314796096.1314800048.10181.gz&fulltext=1