On Wed, Jul 16, 2008 at 09:17, Walter Bender <[EMAIL PROTECTED]> wrote: > It obvious occurred to me after seeing some questions posed on the > OLPC-Sur list: > > 4. There is an inventory of what activities can be used with > mesh, nor what ways.
*no* inventory I presume. I have a list of collaborative activities on http://wiki.laptop.org/go/Collaboration_Central (which hasn't been updated lately) but that is a more developer-oriented page, so it gives details on implementation not features. > 5. The mesh is only capable of some form of collaboration in small groups. The current implementation with the XO's mesh network and telepathy-salut is only capable of small group collaboration - such that other small groups are on separate mesh channels or geographically dispersed so that they cannot see each other. As soon as you have too many XOs on the mesh on one channel, all collaboration fails. The number of XOs depends on the level of activity which can only be determined through much testing. For example, you can have more XOs seeing each others' presence in the Neighborhood view if none of them are running shared activities. Activities that put load on the network reduce the number. Our current scenarios include: * Ad hoc collaboration on the mesh (fully-connected, everyone can see everyone else) * Ad hoc collaboration on the mesh (geographically dispersed, requiring mesh routing) * Ad hoc collaboration on a wireless access point * Server based collaboration (using the jabber server) on the mesh (fully-connected) * Server based collaboration (using the jabber server) on the mesh (dispersed) * Server based collaboration on a wireless access point Release 8.2 may have control panel options to adjust the mesh TTL to help with the fully-connected mesh scenarios. The server scalability work (Gadget) will probably only be available in Release 9.1. Cerebro is being looked at for improving the ad hoc collaboration. There are currently no time frames for its use. > Why don't we add a new field in the Activities page template that > indicates both whether or not an activity supports some form of > collaboration and, if so, what is the supported number of > collaborators. (The latter may, of course, be somewhat fuzzy depending > upon the nature of the connection: via school server or "under a > tree"). > > We could have a simple set of options (the numbers perhaps need tuning): > > A) no collaboration > B) pair-wise collaboration > C) small (3-4) group collaboration > D) classroom (10-20) collaboration > > We could break down collaboration a bit further: > > sharing > interacting > ??? > > and we may want to comment on, for example, how many Type A > collaborations can be supported at once. > > An example of: > A is Turtle Art > B is Distance > C is Write > D is Chat > > We'd need to do some serious QA to figure this out, but I think it > would go a long ways towards giving people a sense of what they can > expect in terms of a robust use of Sugar. We do need to get http://wiki.laptop.org/go/Collaboration_network_testbed going again, to get some idea of the numbers. My understanding is that OLPC's new location for the testbed in a more radio-quiet location will still take some time before it is operational. Since the performance depends on the activity on the entire network, the numbers you suggest would have to be assuming there are no other XOs on the network at all. Regards Morgan _______________________________________________ Sugar mailing list [email protected] http://lists.laptop.org/listinfo/sugar
