"GC" applied to actors is probably a misnomer. A count of the number of live actors is maintained, with live actors being ones that have started but have not exited. Actors that have exited may still be reachable, and can even be restarted. Actors that have not exited can even become unreachable. Pre-2.7.5 Scala Actors handle this by keeping weak references to the actors, so it can be determined when an actor is not longer reachable. 2.7.5 no longer does this due to some suspected poor interaction with the garbage collector when under high loads, but I think the feature is still in trunk.
On Tue, Jul 7, 2009 at 5:39 AM, Xingyang Lu
<luminary.infi@...> wrote:
Hi, all:
I am a college student from The University of Tokyo,
and I am working on some distributed garbage collection algorithm. I
need some well documented testbed of distributed objects run-time
environment for my research. I just wonder if scala is a good choice
for me? And what is the GC algorithm for Scala actor objects? Can
anyone tell me about these issues?
Best regards,
Xingyang
--
路 星洋 (Xingyang Lu)
Chikayama & Taura Laboratory,
Department of Information and Communication Engineering,
The University of Tokyo
Email: luxy@...
--
http://erikengbrecht.blogspot.com/
