browser buster is running for longer and longer these days, but when you get tired and shut down the shut down process can take a very long time. I noticed in the console that I was spinning to shut down over 1500 webshells after running browser buster for a couple of hours on today's linux build. Should we/can we killing off these old webshells earlier? is this a webshell or a frame, or some other problem.

Removing myself from CC and adding Steve (who probably should take the bug).

it's true there are objects that leak webshells. as far as I know, the webshell itself isn't the cause. But I'll accept the bug for now and start looking into it.

Session history is holding onto all the webshells. What's happening is the browser buster works by refreshing a frameset doc every 30 seconds. Session history stores off the frameset (4 frames, ergo 4 webshells) each refresh. Since the frameset is reloaded each time, you actually get new webshells each refresh (vs. having new URL's loaded into existing frames/webshells.) Since session history actually holds onto webshells, all these webshells and the whole cascade of objects they have a strong reference to are held onto until session history is pruned or deleted. Since webshells are heavyweight, it takes a long time to destroy lots of them (I've seen almost 500 one time.) I know travis and radha have talked about how session history should work with the new docshells. I think the design calls for session history to hold onto lightweight objects and not docshells, so this problem should go away. If that isn't the plan, please consider this problem in the design. Assigning to radha.

This s'd be taken care in the new design where docshell will do the loading of urls SH items.

Moving to new Embedding: Docshell component :-)

Moving all to M14 (for webshell changes that are being deferred till M14 rather than destabilize M13).

Moving to M15...

*** Bug 17591 has been marked as a duplicate of this bug. ***

Move to M16 for now ...

*** Bug 32230 has been marked as a duplicate of this bug. ***

This is not valid in the new docshell/SH world.

*** Bug 40233 has been marked as a duplicate of this bug. ***

*** Bug 40233 has been marked as a duplicate of this bug. ***

Verified