On Jan 6, 2010, at 11:11 AM, Mic Bowman wrote: > The interfaces have always existed. They were undocumented & > scattered across files, but you can't have a distributed system > without interfaces. What you've done is make them "less" internal, > slightly better documented, and moved to a single set of files (and > improved them). It's a commitment to a stable set of well-understood > interfaces that makes extensibility possible (or more accurately... > "cost-effective"). Whether they're in an I* file is mostly a matter > of convenience. > > BTW... this is not a complaint about the work that is being done. Or > about the refactoring. This is good stuff. And a commitment to > treating them like real, persistent interfaces is the next step.
Absolutely. You can't do that when your 'furniture' is all over the place, invisible, unmapped, and tangled up with all the wiring in your house. So cleaning up these interfaces and making them publicly visible is the required first step to having everyone commit to their existence as "the contract". This is *it* -- these are the interfaces for resource services that simulators know about. In my mind, the implicit contract with extenders of OpenSim now becomes a reality -- something that wasn't possible when we had bits of the interfaces spread out and mixed up with implementations (unlike what you say, not everything was encapsulated in interfaces, there was a fair amount of direct class instantiations, I can point you to many). You guys did it with a fair amount of pain, and you have A+ programmers in your group. The Rex people also did it at some point, but concluded that it wasn't working; now they're using what your people did. Very few people were able to extend OpenSim given the way that OpenSim was written. Things are completely different from here on, because we made the invisible visible, and cleaned it up in the process. I really don't believe that you believe that OpenSim was ok as it was before... I hope you give this some thought, cos I think that forks are not going to benefit anyone. _______________________________________________ Opensim-dev mailing list [email protected] https://lists.berlios.de/mailman/listinfo/opensim-dev
