On 11/19/09 9:01 AM, Fabio Forno wrote: > On Thu, Nov 19, 2009 at 4:09 PM, Peter Saint-Andre <[email protected]> wrote: > >> If 'ver' for pubsub nodes is conceptually equivalent to 'ver' for >> rosters, then 'ver' refers to the "state" of the node itself (i.e., all >> the items ever published at the node). Therefore it does not exist at >> the item level but at the node level, so 'ver' + ItemID is not a unique >> combination (as you say). However, anytime a new item is published to >> the node then then 'ver' needs to change. Correct? > > Yep, I think that we just just can't say neither that they are unique > nor that they aren't, since they have different scopes. > > Indeed inside the the node 'ver' + ItemID happens to be unique, > because the ItemID is unique and each time we publish something, even > with the same ItemID, the service must produce a new 'ver' for the > node. However I think that saying anything more than any change to the > items of the node (addition, modification, retract) must produce a new > 'ver' is just confusing. > > BTW are there somewhere such things like nodes whose items expire when > they become too old, and is this a deletion?
IMHO that is just a deletion (which might be an automated deletion) Peter -- Peter Saint-Andre https://stpeter.im/
smime.p7s
Description: S/MIME Cryptographic Signature
