[Standards-JIG] JEP-0060: Comments on latest draft.

Fletcher, Boyd C. J9C534 Boyd.Fletcher at je.jfcom.mil
Mon Jun 28 21:12:30 UTC 2004


> -----Original Message-----
> From: Bob Wyman [mailto:bobwyman at pubsub.com] 
> Sent: Monday, June 28, 2004 5:07 PM
> To: Fletcher, Boyd C. J9C534; 'Jabber protocol discussion list'
> Subject: RE: [Standards-JIG] JEP-0060: Comments on latest draft.
> Boyd Fletcher wrote:
> >> 	Append and overwrite=false both require that the server 
> remember all 
> >> itemIDs that have been seen in the past. I can't
> > no they don't. they only are required to know about any currently 
> > published IDs (which they would have to know anyways).
> 	What is a "currently published ID" and why would a 
> server be required to know this "anyways"? Are only thinking 
> about the case of persistent items? Are you thinking that

I would think in both cases a server must keep track of all items currently published. It is only in persistent system does the server keep track of the items after a server restart.
> "currently published" means items that are still in the 
> persistent store and haven't been deleted? If so, are you 
> saying that even if "overwrite=false" it is ok to reuse 
> itemIDs that have fallen out of the persistent store? What do 
> I do with nodes that aren't configured for persistence? What 
> do I do with nodes that store the "last 1,000" messages and 
> don't want to issue a delete notification for every message 
> that drops out of the store? We're writing a spec here that 
> handles a broad range of requirements... It can't be 
> compromised by just one of a number of equally valid ways to 
> run a pubsub server. (i.e. with persistence, without 
> persistence, etc.)

ok then so how would you suggest handling overwriting of items. 

More information about the Standards mailing list