> to do with node/, so it _may_ make sense to run it on a separate port, > and keep low-level FCP simple.
Well, I certainly agree that we should make a distinction between "low level" (or Level 1) FCP (what we have now), and higher level FCP (say, with splitfile support, redirect support etc) (lets call this level 2), and then even higher level (level 3) with FEC support with healing, and possibly functionality similar to that used by Frost and FMB. One way to do this would be to run them on separate ports, but another would be to simply make the distinction explicit in the Spec. If we are clever about it, Level 1 should expose a Java API with a 1-1 mapping between each level 1 FEC command, and a Java method. Level 2 should be built on-top of this API, while exposing its own, and Level 3 should be built on top of the Level 1 and Level 2 APIs. This means that to replace the node implementation, we would only need to reimplement the Level 1 API, Levels 2 and 3 would follow. If a third-party created a node in a different language, they could then more easily use Fred's Level 2 and Level 3 implementations. This way we get all the modularity people want, but without the performance and resource overheads of separate VMs. Ian. -- Ian Clarke [EMAIL PROTECTED]|locut.us|cematics.com] Latest Project http://locut.us/ Personal Homepage http://locut.us/ian/
pgp00000.pgp
Description: PGP signature
