[Standards] XEP-0073: Question about service discovery

Rachel Blackman rcb at ceruleanstudios.com
Wed Feb 7 18:25:06 UTC 2007

>> And what about XEP-0073? It does not require XEP-0115 for the basic
>> protocol suite. So a client implementing only the basic protocol  
>> suite
>> will cause lots of flooding (he doesn't know how and what to cache
>> and so sends disco#info and jabber:iq:* requests to anyone).
> So we'll add XEP-0115 to XEP-0073. :P
>> And about jabber:iq:version: Looks like clients will still flood them
>> if they want to display the OS the contact is using.
> Why is the OS any of your business?

Why did we offer it in jabber:iq:version to begin with?

It actually is fundamentally useful for folks in some situations.  If  
I see a friend on via, say, 'libgaim' or 'Psi' for the client info, I  
have no way of knowing if that's on Windows, Mac, Linux, etc.  So  
maybe I go, 'Oh, dude, here's that new build of <some program> you  
wanted me to send you...' and toss the Windows-only executable over  
and he goes, 'Uhm, that's great, but... I'm on my Linux box right  
now.  Can you just send that to me when I'm back on Windows?'

It's a situation that's avoided if I'm aware of which OS they're on.   
(Unless, of course, I ignore that info!)

I agree it's not a requirement, however.  But I'd think we should  
have a way to (optionally) include all the information from  
iq:version in a disco result.  You don't HAVE to include that info,  
but you should have a defined method TO do so if you want.  Then it's  
not a black-and-white matter of 'I must iq:version to get that info'  
or 'I cannot get that info even if it's offered by iq:version because  
IQ version is bad practice.'

Perhaps even more appropriate would be in the caps field of presence,  
optional 'uiversion' and 'platform' tags.

   <c xmlns='http://jabber.org/protocol/caps'

   <c xmlns='http://jabber.org/protocol/caps'
      ext='xhtml ftrans jingle'/>

...they are two pretty different environments, and knowing that I'm  
using the web-based Express version rather than the full native  
version /is/ a useful bit of info for a hypothetical contact in some  
cases.  (Not leastwise, for instance, the fact that if it's embedded  
in a web browser it's not flashing in the taskbar and I'm potentially  
not seeing your messages if I'm buried in another window.)

So while I agree a client shouldn't /have/ to reveal its OS if we get  
rid of jabber:iq:version, I do think the information should be  
optional to offer somewhere else.

Rachel Blackman <rcb at ceruleanstudios.com>
Trillian Messenger - http://www.trillianastra.com/

More information about the Standards mailing list