On 8 August 2012 21:00, Matthew Miller <[email protected]> wrote: > On Aug 8, 2012, at 12:34, Matthew Wild wrote: > >> On 8 August 2012 18:25, Matthew Miller <[email protected]> wrote: >>> On Aug 8, 2012, at 11:09, Jack Moffitt wrote: >>>> It would also be very nice to formalize session attachment somehow, >>>> but this is probably best done in a new XEP. >>> >>> I agree session attachment probably ought to be a new XEP. >> >> Informational? or do you mean for some kind of request-a-session protocol? >> > > It probably is informational; its basically this (if it's what I think it is): > > < http://metajack.im/2009/12/14/fastest-xmpp-sessions-with-http-prebinding/ >
Right - but that is mainly discussing an out-of-band HTTP request with a custom non-standard payload. Standardizing such a payload wouldn't be 'informational' but clearly a protocol. On the other hand if it doesn't specify *how* to do it then it seems little more than a note to say "You can do this!" - which seems like if necessary it could live in the BOSH XEP itself just as easily. Of note - Prosody 0.9 allows the client to include data in the initial session creation request (the possibility of this is not mentioned in the XEP, but it's not forbidden[1] and it doesn't hurt clients that don't use it). Prosody will reply to everything sent in the session creation request in its session creation response. This allows an authenticated session to be established (assuming ANONYMOUS, PLAIN, EXTERNAL or another single-step SASL mechanism) in a single HTTP round-trip using nothing more than standard XMPP. It also allows for other actions to be pipelined too, without requiring server changes (for example in a private client we also include a request to optimistically enable carbons). Regards, Matthew [1]: http://mail.jabber.org/pipermail/standards/2010-September/023809.html (and if you do read the whole thread - yes, I have changed my stance on this since 2010...) PS. By the way: Carbons + MAM turns out to be a nice solution to the cross-tab issue too...
