[Summit] Agenda Items

anders conbere aconbere at gmail.com
Wed May 21 11:45:25 CDT 2008

On Wed, May 21, 2008 at 9:29 AM, Blaine Cook <romeda at gmail.com> wrote:
> Hey all,
> I don't know if/how agenda items are chosen for the summit in advance, or if
> it's pretty flexible (it seemed that way last year, but there it may have
> been the invisible hand of psa guiding discussions ;-)  ), but here are a
> few topics that I'd love to hack on either code or spec wise:
> -- OAuth over XMPP: the use case is enabling client applications of web
> services to authenticate in a user-friendly flow, and use that
> authentication to allow access to certain XMPP resources. I'd love to have
> something firm nailed down, or at least a proposal --- wether or not OAuth
> is involved, the use case is a definite need.

And I think nailing down the language here is important. I've talked a
lot about this in the XMPP community and I think there's a lot of
confusion about what it means to authenticate, to authorize, how it
would play into the web, and how the web might begin to use XMPP
resources if this kind of workflow were nailed down.

> -- PubSub for web services: it would be really interesting for me to sit
> down with people and figure out some best practices around implementing
> PubSub services for web services. Basically the IRL extension of the
> discussion on the social list.

I would love to be involved in this, right now I think there's a lot
of people who know that their use case fits a pub sub pattern, and are
rolling their own solutions (good), but are doing that without a body
of literature and documentation to help them make the bests choices
possible (bad).

> -- PubSub scalability shoot-out: I haven't found good documentation of the
> various Jabber servers' performance characteristics, particularly as regards
> scaling non-c2s or component usages to (i) millions of PubSub nodes, (ii)
> tens or hundreds of thousands of subscribers per PubSub nodes, and (iii)
> tens of billions of messages delivered per month. I'd be very interested to
> know more about what the various servers are capable of.
> -- XMPP as an internal messaging platform: I ended up rolling my own
> messaging server for Twitter (Starling), but I'm curious to know what level
> of scale XMPP has seen as a messaging bus for internal systems. There are a
> few really compelling arguments for doing so (simplicity, consistency), and
> a few really compelling arguments against (complexity, performance), but it
> would be fantastic to get the real goods on this.

I think this is huge, and could result in some very interesting XEPs.
Right now XMPP is /nearly/ at a point where it can be used in place of
systems like MQ or MSMQ (especially if reliability is less of an issue
than the message passing). Getting this use case nailed down, and
understanding what the current limitations are would be awesome.

> b.

More information about the Summit mailing list