[plt-scheme] threads, memory leakage?

I don't know the details of either the 3m or the Boehm GC,
but I do know that in general GCs don't collect until the
heap is exhausted (this is more efficient).  Therefore, my
first suggestion is to establish that you do indeed have a
leak, either by forcing the process to run out of memory,
or by observing steadily increasing memory usage when you
trigger the GC (using (collect-garbage)).

If you do have a leak I don't really have any suggestions
beyond using 3m, or restarting every once in a while.  I
think 3m has some features to track memory 'ownership', but
I have never used them.

N.

--- Hans Oesterholt-Dijkema <hdnews at gawab.com> wrote:

> 
> >As for the memory usage, 114MB isn't that much. 
> 
> At the end of the evening it was up to 165MB. It just
> increases
> about 2MB per test run. It isn't that the server side is
> very
> busy or so. And you know, I'm developing for people with
> 256MB machines. For those machines, 160MB is much.
> Having GC working on swapped out memory isn't very nice
> I've heard.
> 
> Btw. I don't know if the sqlid drivers I'm using in the
> server side
> backend will work that well with 3m, although that may be
> quite easily to fix. I'm pretty sure though, that the
> mzgtk2 binding
> to Gtk+ (v2) that I'm using on the client side will not
> be usable with
> m3 at the moment, without a lot of work involved.
> 
> What's more: I don't see increasing memory that much for
> mzgtk2 applications. What would you advice in this
> context?
> 
> --Hans

Email: noelwelsh <at> yahoo <dot> com   noel <at> untyped <dot> com
AIM: noelhwelsh
Blogs: http://monospaced.blogspot.com/  http://www.untyped.com/untyping/

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com