From: Joshua J. B. <con...@co...> - 2005-01-28 15:34:55
|
On Friday 28 January 2005 01:41, Martin C.Atkins wrote: > Hi, > > On Thu, 27 Jan 2005 23:29:17 +0100 Mads Kristensen <md...@un...> wro= te: > >.. > > So I was thinking: How about implementing a small FUSE filesystem called > > RepFS (for replicated-filesystem) that replicates all data in two > > Reading this, a thought occurred to me. > > How about adding: > > ...and if the secondary directory is not reachable, logging > the changes to a file. > > When the secondary directory becomes available, replay the logfile > to bring it up-to-date with the primary. What happens if changes have occurred on the server in the meantime? (thin= k=20 of a CVS conflict when trying to commit) > If we had this, we would have a primitive but usable mechanism for > dealing with sometimes connected, sometimes disconnected clients, > with a reasonably efficient and automatic re-sync mechanism upon > re-connection. > > The difficult part seems to be suitable definitions and implementations > of "not reachable" and "becomes reachable". You also have to worry about this synchronization issue. How do you handle= =20 the scenario when the same files have changed on the client and server, but= =20 in different ways? You need to merge them somehow ... or throw it back to= =20 the user and say "here, fix this". :) [...] > I think this could be very useful! I agree, I would love to see an alternative to Coda (and InterMezzo). But = I=20 think such a thing will necessarily be pretty involved, because of this=20 conflict/merging issue... =2D- Josh =2D-=20 Joshua J. Berry "I haven't lost my mind -- it's backed up on tape somewhere." -- /usr/games/fortune |