In general using STM for shared state concurrency in everything is mutable languages like python seems to be destined for failure. Haskell and Clojure both limit mutability and can use STM effectively. The C# team put extraordinary effort into doing a STM for the CLR and failed miserably.
It looks like we need to either port the object level locking capability of the JVM and CLR to python to kill the GIL or python needed new semantics around mutability. US number: +1 650-492-8286 Indian Number:+919945111824 On Sat, Jul 2, 2011 at 11:17 AM, Senthil Kumaran <sent...@uthcode.com>wrote: > On Fri, Jul 1, 2011 at 2:15 AM, Dhananjay Nene <dhananjay.n...@gmail.com> > wrote: > > > Apparently PyPy is going to lose the GIL and implement locking via STM > > > > > The world of databases immediately striked me, as soon as I read the > word 'transaction'. STM seems a good extension to that much testing > concept to a much lower level at the translation stage. Interesting. > > -- > Senthil > _______________________________________________ > BangPypers mailing list > BangPypers@python.org > http://mail.python.org/mailman/listinfo/bangpypers > _______________________________________________ BangPypers mailing list BangPypers@python.org http://mail.python.org/mailman/listinfo/bangpypers
