On Thu, Sep 1, 2011 at 9:12 PM, David Soulayrol <david.soulayrol at gmail.com> wrote: > Hello. > > I've got questions about memory pools management and I have found on > the list a related question which was not answered as far as I can > tell: > > http://article.gmane.org/gmane.comp.voip.pjsip/13275/match=pj_pool_release > > Here is a simple example. I have a hashtable which is instantiated > from the very beginning of my program execution and which lasts until > it ends. Many objects are added and then removed in this table. I need > a pool to insert elements in this table, but I cannot use a global > pool for all of them because it could lead to memory exhaustion. The > solution I have found is to associate a pool to each entry in the > table. So memory is freed when expired elements are removed from the > table and destroyed. Does it sound good? What's your solution? > That's exactly what we do. > More generally, how do you handle the problem of memory management > when there can be unlimited successive allocations of objects with a > long lifetime, like in the pjlib's balanced tree or hashtable ? Didn't you answer this yourself above? > Do you > also use good ol' malloc at the same time? > Using malloc would be the same as admitting defeat, wouldn't it. So no, no malloc. :) Benny > Thanks. > -- > David > > _______________________________________________
