🚀 go-pugleaf

In article <1378@frf.omron.co.jp>, andrew@lsa.ncl.omron.co.jp (Andrew Walduck) writes:
|> Having used this package as a student...I strongly recommend that anyone
|> who is interested in using concurrent algorithms take a very serious
|> look at it.
|> 
|> Its very slick.
|> Image tasks, threads, and co-routines as C++ objects. To send, or receive
|> messages from them, just call a member function...
|>

So how does this package differ from the AT&T task library?   Can you
tell us more about the multi-processor ports? ie. are tasks mapped to
processors automatically or does the programmer have to do that?
 
|> Imagine spawning a task with:
|> 
|> uTask ptr = new aTask(parameters....);
|> 
|> communicating with it with...
|> 
|> ptr->aMessage(someMessage);
|> ptr->aDifferentMessage(anotherMessage);
|> 
|> and waiting for it to complete with...
|> delete ptr;
|>

You can do all this with the AT&T task library.

The initial posting seems to indicate that uC++ is an extension to
C++, and not just a set of classes.   The AT&T task library is simply
a set of base classes which one can inherit from to obtain co-routines.
This seems to be more desirable than augmenting C++ to arrive at yet
another language. 

The multiprocessor ports sound interesting though.  Anybody know if
the AT&T task library has been ported to any multiprocessing
environments?
 
|> 
|> Cheers.
|> Andrew
|> 
|> P.S. How's this for a glowing testimonial??? ;-) I liked using it...
|> 
|> Andrew Walduck -- OMRON Japan --andrew@lsa.ncl.omron.co.jp

-------------------------------------------------------------------------
Jonathan R. Engelsma                            Michigan State University
                                                     Department of
E-mail:   engelsma@cps.msu.edu                     Computer Science
	  uunet!frith!engelsma	
	  uunet!frith!jresys!engelsma (home)	
-------------------------------------------------------------------------