[Standards-JIG] the great infobits reversion

Ralph Meijer jabber.org at ralphm.ik.nu
Wed Feb 25 19:56:36 UTC 2004

On Tue, Feb 24, 2004 at 03:49:03PM +0100, CORVOYSIER David FTRD/DMI/REN wrote:
> Wow ... Quite a reversion indeed !
> I've read with great interest your discussion with Ralph and Jacek, and
> I must admit that though I found at first the Infobits concept
> attractive, I later realized that it was weaker than the already
> existing XML mechanisms (I'd better be studying a little bit more the
> XML concepts ...). I then think you were right to do the "infobits
> reversion".
> However, two questions raised during your discussion remain:
> 1 - Should we define a rule to explicitly say that existing XML schemas
> (whenever available) should be used instead of custom Jabber datatypes
> when defining Jabber extensions (this rule already exists in the Jabber
> TAO, but I don't know its status) ?

IMO, you should want such a rule. Sometimes an existing XML schema fits
the intended purpose for the Jabber world, and sometimes it doesn't. What
I do think is that you should /look/ at existing stuff and, if you decide
to roll your own, possibly define a mapping between the existing stuff
and the new stuff.

> 2 - Will Jabber implementations be enforced in the future to provide a
> better support of namespaces to allow easier extensions/conversions to
> other formats ?

I'd certainly hope so. One of the reasons Info bits came into this world
in the first place, is the fact that most implementations cannot handle
namespace prefixes. Current practice is to look for an attribute called
'xmlns' and go from there. Some really bad implementations even always
look for an element named 'x'. Yuck.

What would be really nice is to say for example:

  <iq to="two at example.com" type="result" xmlns:delay="jabber:x:delay" xmlns:mood="http://jabber.org/protocol/mood">
    <stuff xmlns="ralphms:ns">
	    <body>stuff #1</body>
		<delay:x from="one at example.com" stamp="20040225T19:52:00"/>
	    <body>stuff #2</body>
		<delay:x from="other at example.com" stamp="20040225T19:52:03"/>

> Last, Infobits may be dead, I still believe that we should work on vCard
> replacement and extended presence ...




More information about the Standards mailing list