On Mon, 25 Mar 2019 at 12:26, Florian Schmaus <f...@geekplace.eu> wrote:
> On 25.03.19 12:48, Guus der Kinderen wrote: > > XEP-0313 "Message Archive Management" (v0.6.3) relies on XEP-0359 > > "Unique and Stable Stanza IDs" to identify content in the archive. > > > > MAM provides an archive on the server, which can be efficiently > > synchronized with a client-sided archive. It does this using the IDs > > from XEP-0359. It's a simple query: the client provides an ID of a > > message in the archive, the server responds with all later messages. > > > > Although this is a simple, elegant solution, it breaks when the client > > receives messages that have XEP-0359 IDs, but are not part of the > > (server-sided) archive. This puts quite a restrication on XEP-0359 use > > in other contexts than MAM. Can this be improved upon? > > I think some more context could be helpful. As far as I know this come > up yesterday (?) in the xsf@ MUC in a discussion between Guus and MattJ > (and waqas). > > As far as I understood it, prior MAM versions used an <archived/> > element which not only contains the ID of the message stanza in the > archive but also carries the semantic that the message was actually > archived. > > Now Guus wants for some reason to slap a <stanza-id/> on every message, > irregardless if it was archived or not. MattJ argued that this would > break client's assumption that messages with a <stanza-id/> are actually > archived. This assumes that clients simply remember the last stanza-id > of the message they received and perform a resync of the archive when > they come online again. > > Keep in mind that MAM archives may not archive every message stanza, > think of IBB messages and the like. > > I do not claim to understand every statement and argument in this > discussion, so please correct me if I am wrong. I am also not sure if > the problem is an actual problem: There was a interesting discussion > between MattJ and waqas regarding the guarantees client developers can > expect from a MAM archive. > > But what I understood is when XEP-MAM switched from <archived/> to > <stanza-id/> there possibly, depending on the guarantees you expect from > MAM, was a loss of semantic. > > I see multiple potential solutions: > > 1. Introduce an archived flag > 2. Use <stanza-id/>'s 'by' attribute to carry the archived or not semantic > 3. Potential others > > I lean towards 2. Which could mean that > > <message from='f...@bar.com' to='u...@example.com'> > <stanza-id by='example.com' …/> > </message> > > does not indicate archival of the message into u...@example.com's MAM > archive, while > > <message from='f...@bar.com' to='u...@example.com'> > <stanza-id by='u...@example.com' …/> > </message> > > does. > > Ah, so the first suggests the message is archived by the server-level archive? > - Florian > > _______________________________________________ > Standards mailing list > Info: https://mail.jabber.org/mailman/listinfo/standards > Unsubscribe: standards-unsubscr...@xmpp.org > _______________________________________________ >
_______________________________________________ Standards mailing list Info: https://mail.jabber.org/mailman/listinfo/standards Unsubscribe: standards-unsubscr...@xmpp.org _______________________________________________