Jython, GILs and object locking.

[Daniel Dittmar]

Harri Pesonen wrote:
> You are probably asking next "how threads can communicate". They can 
> communicate with messages, and via special shared interpreter state, 
> which needs a new "shared" keyword in python. Access to these shared 
> variables is synchronized, all other variables are local to thread, and 
> they need not be synchronized.

So you're not talking about Python threads in the meaning of the module 
'thread'. Instead you want to be able to create multiple interpreters in 
the same process and run them in parallel via threads. Today, the Global 
in GIL is really global instead of interpreter local.

I'm not saying that this isn't a worthy goal. Server applications with 
embedded Python interpreters would profit from it and it is not that 
difficult to implement. (Although keeping the patches current with 
Python development will be awful as adding the interpreter state will 
affect every call at the C level).

I could even imagine that extension methods get an addditional flag that 
signals that this methods is able to run in the GIL free environment. 
Older extensions don't have that flag, so Python has to suspend all 
other threads  while this extension is running. This is perhaps a way to 
achieve backwards compatibility.

Daniel