On Tue, Mar 18, 2008 at 10:15 AM, Thomas Heller <[EMAIL PROTECTED]> wrote: > Adam Olsen schrieb: > > > > On Tue, Mar 18, 2008 at 9:21 AM, Steven Bethard > > <[EMAIL PROTECTED]> wrote: > >> On Tue, Mar 18, 2008 at 6:39 AM, Marcin 'Qrczak' Kowalczyk > >> <[EMAIL PROTECTED]> wrote: > >> > Dnia 17-03-2008, Pn o godzinie 11:56 -0600, Adam Olsen pisze: > >> > > >> > > I've replaced __del__ API (which resurrected objects) with a > >> > > __finalize__/__finalizeattrs__ API (which doesn't). Attributes > listed > >> > > in __finalizeattrs__ are proxied into a core object, a finalizer > >> > > thread is given a reference to the core, and when the main object is > >> > > deleted the GC asynchronously notifies the finalizer thread so that > it > >> > > can call core.__finalize__(). The net result is an API very similar > >> > > to __del__ (you need to list attributes it might use), but it's now > >> > > impossible for the GC to run arbitrary code (I even enforce this). > >> > > >> > Ah! Irrespective of other issues, I like this very much. This design > >> > agrees with my understanding of how finalization should behave, except > >> > that I haven't had details in mind which would fit Python. My abstract > >> > design needs the finalization function to somehow access parts of the > >> > dying object, but not the object itself, and this design shows how to > >> > actually do it in a convenient way. > >> > >> Note that you can use something very much like this right now:: > >> > >> http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/519635 > > > > Yeah. That's missing the finalizer thread, but that could be add. > > > > Note, you need to use a finalizer thread in order to use locks to > > protect your datastructures. Without it, if you simply preempt the > > current thread like is done today, you can only do operations the > > memory model provides as atomic. > > > > Hm, I could imagine objects that need to be finalized in the thread that > created them. > How could that be handled?
If they need access to containing objects the only option is to use the underlying tool, deathqueues. WeakKeyDictionary and friends do this. This gives you notification that an object was deleted, but without the core/__finalize__ machinery. This can be wrapped up to give you back the core/__finalize__ machinery, but with you explicitly checking the deathqueue so you can call __finalize__. -- Adam Olsen, aka Rhamphoryncus _______________________________________________ Python-3000 mailing list Python-3000@python.org http://mail.python.org/mailman/listinfo/python-3000 Unsubscribe: http://mail.python.org/mailman/options/python-3000/archive%40mail-archive.com
