"Stephane Le Dorze" <[EMAIL PROTECTED]>: > I need a solution to build a distributed application under heavy loads with > high response time and robustess (MMO). > I have several alternatives; Erlang (Dynamically typed :( ), Haskell (if I > rewrite a distributed layer on top of it but lacks important features), > Scala with its Actors (not ready yet). > I finally found AliceML which seems to provide all what I could ever expect > to build massively distributed application (I have to applause for your > job)!
Hi Stephane, thanks for your interest in Alice ML. It indeed sounds as if it would be a good fit for your problem. However, as much as I would like to say differently: for the time being I cannot honestly recommend Alice ML for critical use on a project like the one you are describing. So far, the Alice system has mainly been a platform for research, experimentation and teaching, maintained by a fairly small team with very limited resources. It is far from being as mature as a system like Erlang or even Scala, especially with respect to concurrent or distributed applications. To answer your specific questions: > - Does any project exist using AliceML in heavy load? No, not that I'm aware of. > - Does AliceML provide all the necessary infrastructure for OTP (Erlang) > like robustess? The Alice library provides fundamental support for distribution, but certainly nothing remotely as mature as Erlang's OTP. There currently is only little for conveniently dealing with failure in distributed applications. > - What are the performance of AliceML? Is the runtime the same as Oz so I > can figure out with the (low level) shootout benchmark? The Alice system is based on somewhat different technology. It should be competitive with Oz on more conventional benchmarks, but definitely hasn't seen the same amount of tuning regarding concurrency and distribution. I would expect Oz to perform superior in these domains. > - How confident could I be that AliceML will still be maintained? Alice is a comparably small project. We very much want to keep it alive, and it will certainly be around for some time to come, but being a research project, we cannot promise anything beyond that. > - Why is there no buzz around AliceML? Maybe because we are somewhat reluctant to aggressive self promotion. :-) > - Did someone already adapt Actor like model in Alice ML? I don't know beyond toys, although it shouldn't be hard. > - Is the garbage collector a stop the world or incremental? It is a stop-the-world generational collector. > - Is there some optimisation for communication bandwidth under the hood? No, nothing worth mentioning. Cheers, - Andreas _______________________________________________ alice-users mailing list [email protected] http://www.ps.uni-sb.de/mailman/listinfo/alice-users
