As far as I can tell, the change in trunk didn't have any effect for me. I'm on Windows 7 with Python 2.7.1 (just using the built-in Rocket server). I added the following code to the top of default.py (outside any functions) in the welcome app: class Foo(object): def __del__(self): pass foo = Foo() foo.payload = [1] * 1000000 I then load the index page of the welcome app and start watching my python.exe process memory in the Windows Task Manager. Every time I reload the page (by hitting F5), the memory increases by about 7MB, which is exactly the same behavior I observe when running the same experiment in 1.91.6 (i.e., before gc was added). These large memory increases do not occur when I remove the above code from default.py. Am I doing something wrong? I'm also wondering if there may be a small memory leak either in the framework or the welcome app, even without the above code. I notice that when I reload the welcome app index page (without the above code), I still get small memory increases, on the order of a few KB per reload. It's not consistent -- sometimes a few KB, sometimes a bit more, sometimes no increase at all -- but memory does consistently creep up as I keep reloading (both in trunk and in 1.91.6). Is that expected behavior? Thanks. Anthony On Friday, January 7, 2011 7:56:51 PM UTC-5, Massimo Di Pierro wrote:
> Mitsuhiko on reddit brings up a good issue that I was not aware of: a > bug in exec. > > code=""" > class Foo(object): > def __del__(self): > pass > foo = Foo() > """ > while True: exec code in {} > > causes a memory leak. Web2py does this. The memory leak is small in > practice and in fact nobody noticed it. Yet it is a concern, in > theory. > > So far I have the current solution > > code=""" > class Foo(object): > def __del__(self): > pass > foo = Foo() > """ > import gc > while True: > exec code in {} > gc.collect() > > It works for me and I cannot reproduce the leak. > I patched trunk with this. Give it a try and let me know what you > think. > > Massimo > >