On Mon, Mar 3, 2008 at 5:15 PM, Fabio Forno <[EMAIL PROTECTED]> wrote: > 2008/3/3 Peter Saint-Andre <[EMAIL PROTECTED]>: > > > We had some discussion about this at the devcon last weekend. E.g. some > > people were interested in extending SXE to be a generalized method for > > synchronizing XML instances (e.g., for DOM events or whatever). The > > conclusion we came to was that defining a protocol for what we called > > "object synchronization" is hard (e.g., conflict resolution is a > > difficult problem), but that it may be easier to define a protocol for > > what we called "event stream synchronization" (just send the change > > notices in order and let the receiving application figure it out). > > More than hard it is impossible to define a general protocol handling > all types of conflicts and synchronization issues. It's an aspect > related to the application logic and therefore all application have > different requirements >
There's nothing impossible about handling the "object level" conflicts (if we consider the object to be just a bunch of key/value pairs). That's what SXE does. However, I agree that it is impossible to see from the transport level whether the collection of the objects as a whole is "inconsistent" from the application's point of view (e.g. they don't map to an XML document that fits the specified schema) but in my opinion the easiest way for the application to solve that problem is by simply changing the objects so that they are consistent! > > > I'm not quite sure what the implications of the devcon discussion are > > for our work on SXE and whiteboarding, but it strikes me that there > > would be consensus for something less ambitious (the XML eventing > > protocol) rather than an advanced protocol for conflict resolution, > > rollbacks, etc. > > What I don't like of the current SXE definition is the inability of > sending the whole document or chunks it, instead of being obliged of > sending all the necessary events for recreating the document. If the extra bandwidth really is such a problem we could define an implicit way of creating the objects out of a chunk of XML as I've mentioned but it doesn't seem worthwhile to me at this point. Joonas
