[Seaside] VM stability on UNIX

Avi Bryant avi.bryant at gmail.com
Mon Oct 24 18:31:45 UTC 2005


On Oct 24, 2005, at 11:23 AM, Lukas Renggli wrote:

>> How much more? Any indication? megs? gigs? gobs?
>>
>
> I deeply tested this half a year ago, ask google for a thread called
> "Strange behaviour of 'WeakArray finalization process'". After the fix
> the average size of a session, within that particular application,
> grows from 5 MB to sometimes more than 40 MB a session (also depending
> on the length of the session). This is no problem if you run your
> application on a machine with a couple GB of RAM, CPU time is more
> expensive and critical to the response-time of the application.

Er... as long as you don't have more than 50 active sessions at once  
(which doesn't seem all that many if they're long lived).
If this is really the only way to reliably deploy (it's not a problem  
I've personally had, but it seems like enough others are), then we  
really need to find a way to reclaim memory within a session, maybe  
even explicitly, that doesn't rely on WeakArray.

Avi


More information about the Seaside mailing list