Memory not being released?

Steve M S Tregidgo smst at quantisci.co.uk
Tue Nov 16 10:57:57 CET 1999


Thanks.  We already have a persistent object system, but I'll take a
look at shelve anyway.

Darrell wrote:
> Here's an idiom I like in this case.
> Store your objects in a dict based on id(obj), just as you've done.
> Now if objects must reference each other use the id(obj)
> and look up the obj. And delete them from the dict when needed.
> Yes it's slow but only use it in these big ugly cyclic data
structures.
> 
> But notice how fast memory gets used up.
> This eats 84meg.
> >>> dict={}
> >>> for x in xrange(1000000):
> ....     dict[x]=(x,x+1)
> ....
> >>>
> 
> You might take a look at shelve. It gives you a persistent
dictionary.
> I've been playing with a flat file manager that can reclaim freed
space.
> 
> --
> --Darrell
> Steve Tregidgo <smst at bigfoot.com> wrote in message
> news:80pdvo$99r$1 at nnrp1.deja.com...
> > Hi there,
> >
> > I'm having problems with a large process, and I'd really
appreciate
> > some help in tracking down the cause of its bloatedness.
> >
> 
> 
> 
> 
> -- 
> http://www.python.org/mailman/listinfo/python-list




More information about the Python-list mailing list