Re: Teaching Concurrency

[manis@cs.ubc.ca (Vincent Manis)]

In article <5598@bd.sei.cmu.edu> firth@sei.cmu.edu (Robert Firth) writes:
>(3) veryslowsort.  the program reads a list of integers and creates
>    one task per integer, whose priority is given by the integer and
>    whose sole action is to print the integer.  the integers are
>    therefore output in task priority order, ie they are sorted.
Aaargh! I feel like I'm flogging a dead horse, posting two articles in
succession on the subject of underlying processor models. 

There is absolutely no reason why this should work. Not all schedulers
guarantee strict ordering of execution on a priority basis. Process
aging is one of many factors which could produce incorrect output.
Similarly, the scheduler might ignore priorities (the multi-task kernel
I wrote, for example!). 

However, veryslowsort is an excellent example of how *not* to use
parallelism. 

Apology dept: My last article contained a slighting remark about Ada. I
didn't notice that comp.lang.ada was included in the Newsgroups: line.
Mindful of Usenet protocol, I apologise for posting this to
comp.lang.ada. 
--
\    Vincent Manis <manis@cs.ubc.ca>      "There is no law that vulgarity and
 \   Department of Computer Science      literary excellence cannot coexist."
 /\  University of British Columbia                        -- A. Trevor Hodge
/  \ Vancouver, BC, Canada V6T 1W5 (604) 228-2394