On Wed, Mar 11, 2009 at 03:57:26PM -0430, Patrick O'Callaghan wrote: > On Wed, 2009-03-11 at 16:42 +0000, Chris G wrote: > > On Wed, Mar 11, 2009 at 11:57:40AM -0430, Patrick O'Callaghan wrote: > > > On Wed, 2009-03-11 at 16:14 +0000, Chris G wrote: > > > > It's trying to do the impossible though. If, for example, one end has > > > > task categories that don't exist at the other end there's nothing that > > > > SyncMl can do to sort it out. > > > > > > Or anything else. How do you propose to deal with this in your > > > home-grown solution? What about when you change phones and the new one > > > isn't exactly the same as the old one? These issues are inherent in the > > > nature of the problem. The best way to deal with them is via a > > > standardized markup language, i.e. SyncML or similar. To take your > > > example: SYncML doesn't define specific task categories, so if your > > > various devices aren't understanding each other the problem is with > > > their use of SyncML, not with the language itself. > > > > > If both ends use the same data file they *can't* disagree! OK, it's a > > little more complex than this but to my mind it's fundamentally less > > broken then trying to bodge it with a 'translator' in the middle. > > The OP talked about syncing two instances of Evo, so it *might* be true > that the data is in the same format (assuming they are both the same > version), but then you introduced the notion of syncing with other > devices such as PDAs, where the data files will definitely *not* be in > the same format. > The idea behind using Webdav/Caldav or whatever is that it *makes* both ends use the same format.
> > SyncMl does 'know' about what it's transferring to some extent, if it > > doesn't then how is it any different from a simple file copying > > mechanism? There are several very clever file synchronisation > > utilities already available, if that was all that was needed then > > SyncMl would be redundant. Where SyncMl scores is that it *does* know > > the sort of things its transferring. > > SyncML is not a file transfer mechanism, or even a synchronization > protocol (despite the name). It's a description language. The only sane > way to interoperate N devices is to have an intermediate standard > format. That way the number of translations is O(N), not O(N^2). That's > what SyncML is for. > Which confirms what I was saying - SyncMl does know (and thus *limits*) what is being transferred. OK, SyncMl can be expanded as needed to allow more different clients to connect using it but ultimately it's a dead end because you have so many possible sorts of data being passed that no two ends ever agreee. Or are you saying that the SyncMl "intermediate standard format" is effectively cast in stone? In that case I see little difference between it and Webdav/Caldav. Using Webdav/Caldav to underpin everything means that essentially you say (in SyncMl parlance) that "these are the things I will transfer" and there are no more. Ultimately you 'encourage' both ends to use .ics format data internally and *that's* when it all becomes relatively easy. -- Chris Green _______________________________________________ Evolution-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/evolution-list
