[Gardeners] Shared State Concurrency specification.

[Marco Monteiro]

Duncan Rose wrote:
> On Sunday, July 16, 2006, at 02:07  pm, Marco Monteiro wrote:
> 
> 1. I don't see any 'join' like functionality; not sure if this might be 
> wanted.

This may be added later. The preferred way to 'join' a thread is to use
a barrier. I would like to hear people on this.

> 2. The document talks about global values of special variables being 
> shared between threads. I don't believe that this is the way OpenMCL 
> works at least, and I'm not sure that sharing these is a good idea (for 
> example, if I start two REPLs and setf *read-base* in one, should the 
> other automatically reflect this change?).

I don't know OpenMCL. The lisp implementations I know share global
values between threads.

> 3. Not sure if it makes any difference whatsoever, but can the 
> functional argument to SPAWN-THREAD be a closure? (I'm not sure this 
> matters much either).

(spawn-thread #'(lambda () (dostuff arg))

is allowed. A lexical closure is a function. Is this what you are asking?

> 
> (Other than that?) the document looks reasonable to me (perhaps I would 
> argue about some function names, but that is rather an irrelevance I 
> think).

Which function names?

> 
> -Duncan
> 
> 

Thank you for your comments.

Marco