[gclist] Another request from a Newbie :-)

[Ken Rose]

"Fahad G." wrote:
> 
> Hi,
> 
> Like you mentioned, there are in fact a few papers available on the
> subject and some ideas and solutions have been proposed. However this
> matter remains un-tackled in that, (almost all) practitioners deny
> (sometimes completely) the idea of garbage collection being introduced
> in a hard real-time system. Some say it's ridiculous and some say it's
> impossible (i.e garbage-collection introduces unpredictability into the
> system which is a no-no in a hard real-time environment). Anyways, the
> point is that you can most certainly choose this topic and maybe try
> convincing the world that there really *is* a way you could introduce
> garbage collection into hard real-time systems.

Take a look at http://citeseer.nj.nec.com/573856.html

It appeared in POPL '03.  The basic idea is that you give the collector
a time slice like everybody else, and that lets you make real-time
guarantees.  I haven't studied it extensively, but on one reading, it
seems entirely workable.

 - ken