Thank you all. I think I get the gist of what happens now.
Really appreciating all of the effort you guys put into helping people on this list! Chris On Sat, Sep 17, 2011 at 4:17 PM, Anthony <abasta...@gmail.com> wrote: > Also, note that your 'results' dict will only persist for the duration of a > single request. Each request is handled by a new thread (from a thread > pool). > Anthony > > On Saturday, September 17, 2011 11:06:14 AM UTC-4, Massimo Di Pierro wrote: >> >> On Sep 17, 6:31 am, Chris Rowson <christop...@gmail.com> wrote: >> > Hi folks, >> > >> > I'm trying to get my head around what happens to data stored in memory >> > when >> > it isn't needed anymore. >> > >> > Let me give you an example. Let's say I create a function which returns >> > a >> > dict called 'results' populated with data from an external source each >> > time >> > a user registers (for instance let's say it has data returned from >> > geocoding >> > the users postcode and providing info about the area). >> > >> > So the 'results' dict holds that data, it is used, and then what? Is the >> > memory automatically freed up again when we finish with the data or does >> > it >> > hang around in memory? >> >> Yes. Python uses reference counting. Every variable is a pointer to a >> memory location. It keeps tracks of how many pointers point to the >> same location. >> >> a = A() # one pointer >> b = a # two pointers >> c = a # three pointers >> >> when you delete one the counter is decreased >> >> b = 4 # now two variables (a,c) point to A() >> >> when the counter is zero the object is freed. >> >> This works great UNLESS the objects have circular references (a.x=a) >> AND the object has a destructor (A.__del__). As long a you do not use >> destructors (who needs them?) everything works great. >> >> >> >> >
