30.08.2010 23:06, Remko Tronçon wrote:
Hi Evgeniy,
Hello :)
IMO, the primary goal of the XSF (and any XMPP software project) is making XMPP users happy. Unable to implement/scale protocols means bad experiences, means unhappy users, so we *definitely* care about implementation feedback.
Good to know :)
Disco-ing clients all the time costs users a lot of money on mobile systems, and makes users very unhappy. PEP's "raison d'être" *is* avoiding all this unwanted traffic at signin.
I don't understand: if you are fighting with unwanted traffic, why do you refuse my suggestion about filtering unsupported stanzas on server using client's CAPS. Since we cache local user's CAPS, we can filter not only PEP, but, for example, unwanted iq:version, iq:time, chatstates, muc invitations, etc.
If it's impossible to implement PEP in a scalable way without sending unwanted traffic, then we should throw this protocol away, and start over with something that *does* work.
Everything is fine from my point of view, except of caching foreign clients data.
But in the meantime, for the sake of the XMPP network, please don't deploy PEP for inter-server communication if this means costing everybody money without *any* means to avoid it.
Well, in the first, we don't have a choice: users want PEP and we need to implement it somehow. In the second, as I said in previous posts, I don't think PEP consumes lots of *useless* s2s traffic.
-- Regards, Evgeniy Khramtsov, ProcessOne. xmpp:x...@jabber.ru. _______________________________________________ JDev mailing list Forum: http://www.jabberforum.org/forumdisplay.php?f=20 Info: http://mail.jabber.org/mailman/listinfo/jdev Unsubscribe: jdev-unsubscr...@jabber.org _______________________________________________
