On Sat, 2007-02-03 at 04:24 -0500, Ivan Krstić wrote: > Joshua N Pritikin wrote: > > Anyway, I was thinking about collaboration. I understand that at least > > Abiword and TamTam have plans and maybe implementations for activity > > specific collaboration. That's great. What I am worried about is apps > > which are not tailored for Sugar. Is Sugar going to provide some sort of > > fallback generic collaboration facilities? > > I believe the plan is to provide not only fallback collaboration > facilities, but recommended ones. Abi will need to move to use the Sugar > presence service, and then do its raw communication over Jabber (the > AbiCollab design supports this already). This lets us do the right thing > when you're on the mesh and when you e.g. come in range of the school > server, or just connect to the net from somewhere.
Right. At the very least, Sugar will provide a mechanism to find people you want to collaborate with. Sugar will also provide a 'bus' type primitive that sends a message to everyone participating in your activity. You will also be able to grab the IP6 address of a buddy as well. But remember; any of the buddies in your activity might be reachable through _either_ the school server, or on your local mesh link. Buddies participating in the activity will likely be a mix of both; so any collaboration code will have to deal with both methods of talking to any single buddy. Sugar will provide an python API to do this easily (maybe a C library later on), but if you don't use it you're on your own for now. Sugar may also provide a higher level API later on that isn't just dumping a blob of data down a pipe to everyone. Like Ivan mentions, there are certainly ways to make the one-to-many communication easier, and those have been discussed here before. Dan > My understanding is that there are two APIs in question here: one is > lower-level, whereby after interfacing with the presence service, you > can go on the wire and speak Jabber directly to implement your > collaboration logic, and another is a higher-level API which lets new > activities use Sugar collaboration primitives without having to roll > their own. Dan and Marco can correct me if I misunderstood this. > > Andrew Clunis posted some ideas about the higher-level API on this list > a bit of time ago, and Alan Kay chimed in with a link to a PhD thesis > that provides a pretty compelling model for collaboration with > shared-state objects. That might eventually be the way to go. > _______________________________________________ Sugar mailing list [email protected] http://mailman.laptop.org/mailman/listinfo/sugar
