[Council] meeting minutes, 2008-10-15

Peter Saint-Andre stpeter at stpeter.im
Mon Oct 27 15:00:52 CDT 2008


Dave Cridland wrote:
> On Thu Oct 23 17:00:43 2008, Kevin Smith wrote:
>> > 11. Proto-XEP: PubSub Chaining
>> > 12. Proto-XEP: PubSub Queueing
>>
>> I'd like to hold off until Dave's comments have come through, but not
>> inherently against them going to experimental.
> 
> Well, here's my "thing". I've not got back to the authors, I'm afraid -
> whenever I've had the time I've not seen them online, and I've code
> freezes and other nightmares.
> 
> On PubSub chaining, I think it's okay, although I suspect unsuitable for
> real repeaters - I think this model is right for aggregation, but not
> terribly interesting for the simple case. I've another design for
> shadowing entire services, both MUC and PubSub, that I really need the
> time to finish off.

Right, I think the use case Ralph had in mind was really aggregation.
Indeed I'm not really sure about that, but presumably he can illuminate
the matter. :)

> On PubSub queueing, I'm concerned that this is entirely the wrong model.
> In particular, I'm wondering what happens when a na�ve pubsub client (or
> a repeater, like above) subscribed with the default options, expecting
> one behaviour and discovering another entirely. I suspect - but I'm
> really not sure - that it might need to be an entirely distinct interface.

I've forwarded your comments on to Joe and perhaps we can have a
discussion about this on the pubsub@ list.

Peter


More information about the Council mailing list