On Mon, Aug 01, 2005 at 08:46:07 -0700, Greg Buchholz wrote:

>     Has any thought been given to using a concurrent garbage collector
> for Perl6?  Besides eliminating GC pauses (which in turn means less of a
> need for users to fiddle with the GC settings, and therefore a smaller
> chance of accidently screwing it up), it might be one of the easier
> sources of parallelism to exploit on multi-processor machines.  And it
> could open the door for more soft realtime applications in Perl (audio
> processing, games, etc.). 

Well, the reason an API based on what an object requires, not how
it gets it is better is that it doesn't matter - the GC could be a
separate thread, and GC::timely could just be a blocking call to the
GC thread asking it to check of the said object needs to be
collected or not. This could be optimized for latency, so that just
the critical objects are checked first, and then everything is
cleaned up in the background.

-- 
 ()  Yuval Kogman <[EMAIL PROTECTED]> 0xEBD27418  perl hacker &
 /\  kung foo master: /me beats up some cheese: neeyah!!!!!!!!!!!!!!!!!

Attachment: pgpFCwFyVlXJl.pgp
Description: PGP signature

Reply via email to