-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Old thread alert!
On 7/7/09 7:29 AM, Dirk Meyer wrote: > Peter Saint-Andre wrote: >> On 6/3/09 9:43 AM, Dirk Meyer wrote: >>> Brian Cully wrote: >>>> I'm not sure if it's entirely appropriate to shoehorn it into >>>> pubsub, since you can probably take care of this within a given >>>> application, but doing it at the pubsub level would make it easier >>>> for app developers and allow for easier distribution of item >>>> publishes without having to share state outside of XMPP. >>> It works without extra support in pubsub. It would be similar to roster: >>> on every startup my application would contact all pubsub servers it >>> stores stuff on and fetches all items from the persistent storage. My >>> idea would reduce the traffic just like roster versioning does. It is a >>> nice add-on. >> What exactly needs to change in the spec to make this happen? > > Bob Wyman gave me the idea to re-use the timestamps discussed on another > thread. The idea is good and solves half my problem. Using the timestamp > I can keep track of changes on the server. What's missing is a way to > use auto-subscribe with that feature. I want to know if the node has new > or deleted items since I was last (auto) subscribed. Instead of sending > the latest item on subscribe, the server sends the last timestamp of a > change on the node (which could be a delete event that can not be > detected by sending the latest item). > > So what I need as spec besides the timestamp discussed in the other > thread are a two new possible values for the configuration option > pubsub#send_last_published_item: timestamp_on_sub and > timestamp_on_sub_and_presence. The behave similar to the values without > the timestamp_ prefix. The difference is that the server does not send > the real last published item and only the timestamp of the last change > instead: > > <message from='pubsub.shakespeare.lit' to='[email protected]' > id='autosub'> > <event xmlns='http://jabber.org/protocol/pubsub#event'> > <items node='princely_musings' timestamp='...'> > </event> > </message> > > Does this make sense? As far as I can see from the list discussion, we did not have consensus on making this change. Please correct me if I'm wrong about that. Peter - -- Peter Saint-Andre https://stpeter.im/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.8 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAkrCmTgACgkQNL8k5A2w/vxHtQCdEwGE42MEqu/DBomPQ8VEGFSY drYAnAix8uS+3qfgdELfHxOYZyt3Z3UF =G70a -----END PGP SIGNATURE-----
