There is nothing wrong with the in-process (local) services, and I think they should stay. They're as much services as the remote ones, and are quite handy for standalone grids. I hate having to track N different processes.
This is all a matter of finding the right *vocabulary*, which usually means having a good model in place for what's going on, and the right mechanism to support it. If we start looking at OpenSim as a sort of service-oriented architecture, then it's clear that the issue of which processes host which services is a configuration issue, not an architectural one. Justin Clark-Casey wrote: > Tommi Laukkanen wrote: >> Well said Michael and I have come to similar conclusions when working >> with OpenSim code. There is no need to have separate code for grid and >> standalone. You can run the systems using separate executables like >> standalone and distributed services and still keep the code base same. >> Separate executables are just a question how you deploy and fire up >> your system. >> >> I might go even further and make decoupled services discuss over >> network even if they are hosted in the same sandbox process. It is not >> perfect for sandbox performance but then you would be running the same >> code in local testing setup and in production grid. I would say this >> is a big quality boost. It would also make everyone more concentrated >> on the remote API performance as those APIs would then be used in >> sandbox mode as well which is the environment mostly used on >> development time testing. > > I would just note that we've been here before in list history and a > significant number of us want to retain the existing > in-process services for various use cases. Look through the mailing list for > the previous discussion. > > That being said, if someone wants to produce a 'remote API' based single > sandbox executable then please go ahead. I'm > sure that if it's demonstrated to work just as well as the in-process > services then fewer people would object to switching. > _______________________________________________ Opensim-dev mailing list [email protected] https://lists.berlios.de/mailman/listinfo/opensim-dev
