Comments on: IE’s Memory Leak Fix Greatly Exaggerated

By: tomh

tomh — Fri, 16 Oct 2009 05:42:40 +0000

i know this post is old, but just in case anyone want to reference the article mentioned in the post, it’s been relocated here: http://blog.j15r.com/2007/09/ies-memory-leak-fix-greatly-exaggerated.html

By: Soney

Soney — Tue, 29 Jan 2008 18:52:11 +0000

Has anyone faced an issue of leaks when using prototype.js and dojo.js in the same page ?
For this leaks like anything in IE 6 & 7 :( !!!

By: Nikita Kapustin

Nikita Kapustin — Mon, 24 Sep 2007 01:17:25 +0000

Oh, come on, http://browsehappy.com/why/

Be happy, now ;)

By: Joel Webber

Joel Webber — Thu, 20 Sep 2007 14:44:41 +0000

We also do some pretty awful backflips to work around this problem, though they tend to remain buried in the low-level GWT libraries. The hardest problem to deal with, in my experience, is avoiding leaks *while the app is running*. That is, if I create a widget, use it, then throw it away while the app is running, it needs to get cleaned up.

Otherwise, you’ll end up blowing enormous amounts of memory in a long-running app.

This article’s very GWT-specific, but will give you some of the flavor of how we approach the problem: http://code.google.com/p/google-web-toolkit/wiki/DomEventsAndMemoryLeaks

By: Frank Thuerigen

Frank Thuerigen — Wed, 19 Sep 2007 18:59:35 +0000

IÂ´m so happy I saved my time throwing my own related code out of my lib… I really donÂ´t understand this behaviour from Microsoft. We really need cross-browser JS engine plugins with auto-update and a valid open source reference implementation… :-(

By: Frank

Frank — Wed, 19 Sep 2007 15:21:05 +0000

Alex,

Can you elaborate a bit on what you said about “For things to be cleaned up correctly, we have to ensure that things donâ€™t actually get destoryed, but rather added to the visible DOM just before unload (as the pattern describes would imply)”?

One thing I would love to see is a resource which explicitly describes the leak patterns, and proper ways to deal with them.

By: Andy

Andy — Wed, 19 Sep 2007 04:32:19 +0000

Wow, what happened to automatic garbage collection?

By: Bertrand Le Roy

Bertrand Le Roy — Wed, 19 Sep 2007 04:15:43 +0000

Told you so :p
http://weblogs.asp.net/bleroy/archive/2007/06/26/ie6-s-main-memory-leak-hot-fixed.aspx
We didn’t change a thing in ASP.NET Ajax after this hotfix, which is not fixing anything for all practical purposes. As Alex says, the cleanup code is pretty much still needed in every toolkit.

By: Alex Russell

Alex Russell — Tue, 18 Sep 2007 22:56:10 +0000

we got lucky in Dojo 0.9 that we were testing for memory leaks and spotted this early. We re-enabled the memory leak prevention code just before release for IE 7…we’re still finding new leakage patterns. For things to be cleaned up correctly, we have to ensure that things don’t actually get destoryed, but rather added to the visible DOM just before unload (as the pattern describes would imply). Real DOM cleaning code still has to be part of every toolkit. It’s effing insane that we’re still doing this shit in 2007. IE’s COM dependency is going to be the death of us all.

By: Trevor

Trevor — Tue, 18 Sep 2007 21:19:16 +0000

Safari also has memory leaks, and I doubt Opera doesn’t. The problem though is that IE’s developed in an environment that precludes community help improving the browser.

By: Georg Miller

Georg Miller — Tue, 18 Sep 2007 19:12:40 +0000

IE is not the only one that has memory leak problem. If you watch ff memory usage while interacting with ajax app (eg. ExtJS) you will also see that the usage is increasing. It does release some of used memory, but apparently some are still stuck and can’t be released.

By: scriptkiddie

scriptkiddie — Tue, 18 Sep 2007 18:29:24 +0000

Only loosers still use IE nowadays

Enjoy the web, there is plenty to do if your browser let you

By: Stephen Kellett

Stephen Kellett — Tue, 18 Sep 2007 15:50:33 +0000

Hmmm, We wanted to support IE with Java Script Memory Validator but the IE team members change so rapidly (being pulled off one thing onto another) that the people inside Microsoft that were trying to help us could no longer make it happen. If Microsoft would talk to us we could have a working memory tool for IE in weeks. Stephen