> > I think you can apply real-time scheduling theory here. 
> AFAIK (provided
> > that the tasks are scheduleable!) earliest-deadline-first works OK.
> 
> I think real-time scheduling (the way it's implemented in 
> RTOSes) is too 
> simplistic for what I'm trying to accomplish. Some tasks 
> *should* starve 
> because in the overall scheme of things other tasks are more 
> important.

There are RT-scheduling strategies that implement all kinds of
tradeoffs. But when you need one of those you have first to settle
yourself on what you realy want. No software can solve an unsolveable
problem. If your problems realy have fixed deadlines, and they can be
completed in time, earliest-deadline-first will work. If not, you don't
have a solveable hard-real-time problem, and you will have to provide
more details.

Earliest-deadline-first is a good scheduling when granularity (= task
switching overhead) is high. 

If the realation between deadline-overruns and consequences (money) is
linear you maybe could use linear programming.

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: www.voti.nl/hvu
 

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist