On Wed, 2011-08-24 at 17:04 -0600, Peter Saint-Andre wrote:
> On 8/24/11 4:52 PM, Matthew A. Miller wrote:
> > Having the ability to know when a vCard changes without having to 
> > poll is very very very nice.
> 
> Agreed.
> 
> One stumbling block, to me, is the publishing side. How does a client
> know which publishing method is preferred? Do we allow only PEP publish
> or IQ publish, and expect the server to make the information available
> to both kinds of subscribers/retrievers?

Transparently mapping IQ set/gets to equivalent PEP publish/retrieve on
the server-side doesn't sound so hard. If both ways are available and
does the same, then what difference would it make? So, how about leaving
preferences up to implementors?

If we do go with allowing both, then all servers with PEP support
already support half of it. That instantly gives clients with PEP
support something to test against, which may speed implementation and
adoption. And that should encourage server devs to implement the other
part.

I'd expect the IQ way to be used when PEP is missing (like a client that
didn't already support it but really wants to have vCard4) or unusable
(like in MUC rooms). And maybe if you are cheap and want to save a few
bytes. ;)

Also, maybe we should also look at making XEP-84 MUC-friendlier?
-- 
Kim Alvefur <[email protected]>

Reply via email to