Re: [Axiom-developer] Re: Some code, if you are interested.

[Stephen Wilson]

root <address@hidden> writes:
> I strongly recommend reviewing the clojure for lisp videos by Bill Hickey. I
> think he has some fundamentally useful ideas for exploiting parallelism which
> could be reflected into Axiom.

Just finished watching the videos.  Great talker, great ideas.

Of course now I need to sit down with Clojure and experiment.


A thread or concurrent domain could be written using any number of primitives
various lisps offer (posix thread model on SBCL, fork/waitpid model in GCL,
etc).

In many ways using the fork/waitpid model smooths away some of the headaches
having mutable data can bring since the child always gets a copy of the heap and
stack.


Sometime soon I should find the time to implement a simple thread domain using
my little compiler to see what happens.  I kept concurrency in mind when I was
writing the thing so the system should (I hope) work in a mutithreaded world --
there are only a few points of possible contention as far as the runtime is
concerned.

We could implement a traditional mutex/condition-var model and see how it
"feels" (probably just the same as it does everywhere else).  And since, at this
time, the system has no library and only one destructive primitive (which could
be trivially removed) we could implement Hickey's functional datatypes and
explore the methods needed to get STM.

I would not go so far to say the above would be easy, but it might be a
comparatively expedient way to help understand concurrency in the context of a
spad-like language and to grow familiar with the implementation details.

Steve