Concurrency in Icon has been studied; Kelvin Nilsen's Ph.D. dissertation
was on an experimental system (Conicon, concurrent Icon) which was never
released.

Concurrency in Unicon as Michael proposes might be an excellent addition.
Here are a few additional thoughts on his proposal.

1. making the VM and runtime system (including the garbage collector)
   thread-safe is a potentially Huge job.  Clever techniques and
   compromises might enable one to avoid a massive rewrite.

2. co-expressions are indeed synchronous, co-operatively scheduled
   threads.  One option for pseudo-concurrency would be to extend
   the existing co-expression mechanism to include a scheduler and
   preemptively switch threads every so often under explicit control
   of the VM.  Problem #1 only comes up if you want true concurrency
   on a true multiprocessor, or are building atop the Posix (or some
   similar) threads package.

3. Since Java is thread-friendly, if you need true concurrency
   instead of pseudo-concurrency, it might be easier to modify Jcon
   to include a threads facility, and to modify the Unicon translator
   to work with jcont, than modify the Unicon VM.  I am not pushing
   this, I just wanted to point it out.

4. Both the Unicon translator and underlying icont translator currently
   use YACC but not Lex; they have handwritten scanners.  This does not
   contradict Michael's assertion that it would be fairly "easy" to
   extend the translators to include new syntax for threads.  I think
   the big hard job would be #1 above.

Cheers,
Clint


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
Unicon-group mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/unicon-group

Reply via email to