Re: [Jcyclone-users] Jcyclone - RIP ?
Brought to you by:
jm7
|
From: Toli K. <to...@ma...> - 2006-10-30 19:46:23
|
Anthony, We've been using JCyclone for about 6 months now, and it does a great job of fulfilling our needs. We are developing an open-source financial trading platform (http://www.marketcetera.org), and we use JCyclone as the communication backbone for routing orders through the different stages of the system. JCyclone is 100% pure java and thus OS-independent. We have it running on Windows, Linux and MacOSX without any issues. We found the project to be fairly complete and stable, and we fixed the bugs that we found. We've implemented a new scheduler based on java.util.concurrency classes which provides a basic implementation of the system using built-in concurrency. As far as other projects, there's MuleSource (http://www.mulesource.org) which implements JCyclone features as a subset of a much larger and more complex enterprise-service bus platform. Hope this helps. Since we are both writing financial products, I'd be very interested in your findings and your evaluation of this space as well. toli On 10/30/06, Mark Martin <sto...@gm...> wrote: > Anthony, > > A few answers... > > A) I can't speak for Matt Welsh, but there certainly are a few contributors > still contributing.. As near as I can tell, Jean Morisette is the official > maintainer, and Toli Kuznets has been contributing a lot over the last few > months. I haven't yet made any official contributions, but I do plan to. > Soon. > > B) > 1) Yes. I have a project (not yet released) running on it. I know Toli is > doing some interesting things with it. YMMV, of course. I haven't seen any > large scale testing on it, but I have a test framework that could probably > be used as a genesis for such testing. > 2) No. Although Toli recently acheived concensus that this should probably > happen -- as well as a move to JDK 1.5. One of the things I plan to > contribute to. > 3) Yes -- it's pure java. I have it running on at least Solaris and Win > Xp. There are no JNI wrappers or anythig like that -- and I've got native > threads running on the Sparc T1 so I'm happy. It is, of course, subject to > inherent Java bugs, including one prohibiting loopback TCP connections on > Solaris (fixed in Mustang). > > C) I found nothing like it when I was searching for such a framework. My > guess is you'll spend months trying to replicate current functionality in > any new project. > > > > On 10/30/06, Warden, Anthony <Aw...@le...> wrote: > > > > Hello > > > > I don't see any activity on Jcyclone for the last 6 months - does > > someone know if the project is A) Dead B) Completely stable and feature > > complete C) Superseded > > > > For A - where did the contributors go ? > > > > In the case of C - what's the related effort ? > > > > For B -> > > 1) Does anyone know if the SEDA feature of adaptive admission control > > (per > http://www.eecs.harvard.edu/~mdw/papers/control-usits03.pdf > ) is a > > feature in Jcyclone? > > 2) Is the Jcyclone code implemented over the java.util.concurrent > > classes (and should it be) or are all the fundamental's implemented > > from scratch - e.g. > org.jcyclone.core.queue.IBlockingQueue vs > > java.util.concurrent.LinkedBlockingQueue > > 3) Is the code free of any OS or specific JVM or native code > > dependencies > > > > > > Thanks for your help. > > > > Anthony > > > > > * Anthony Warden > > > * Electronic Client Services Technology > > > * Lehman Brothers International (Europe) > > > * 25 Bank Street, London E14 5LE > > > * aw...@le... -- Toli Kuznets http://www.marketcetera.com: Open-Source Trading Platform download.run.trade. |