[Standards-JIG] JEP 60: Deleting Nodes (8.2.3)

Peter Saint-Andre stpeter at jabber.org
Mon May 23 16:57:25 UTC 2005


On Sat, Apr 23, 2005 at 12:41:29PM +0200, Ralph Meijer wrote:
> On Fri, Apr 22, 2005 at 03:42:43PM -0700, Chris Mullins wrote:
> > [..]
> >
> > I've personally implemented a lot of jeps at this point (that would be
> > an amusing stat - who on this list has implemented the most JEPS? Heh)
> > on both the client and server side, and given the frequency with which
> > I'm missing some of these little details - like affiliation JID's are
> > bare JID's - makes me nervous as to the compatability between the
> > various implementations. 
> 
> Very true.
> 
> This specification is pretty large, so it is hard to get the whole picture
> anyway. I've also discovered that the current text has also proven to be very
> initimidating to people. First of all, people find it hard to 'get' the idea
> behind the publish-subscribe design pattern. Second, when reading the text,
> people think it is too difficult to use, while the basics are actually quite
> simple.
> 
> I think we at least need to do the following before moving to final:
> 
>   - Spell out more clearly what part of the specification is mandatory and
>     what is not. Maybe we need to abandon the current structure, and divide
>     the text into a first part, with all the basic stuff. And then add to that
>     in later parts.
>   - Exactly specify which disco features cover what parts of the
>     specification and their interaction with other parts.

Sounds like a job for the JEP Editor. :-)

I agree that the text as currently written is hard to grok. That would
suggest the need for some tweaking or perhaps even a reorganization.
It's no fun to reorganize a big JEP like this (refactoring even a
relatively small JEP like Data Forms was a big job [1]), but if it is 
necessary then we'll need to bite the bullet...

> Also, I suppose some tutorial like prose would be nice, but not in the
> specification. I've been thinking about starting this, as a follow-up to
> the presentation I did at FOSDEM.

Well, we can always add a non-normative section that provides a better
introduction to the publish-subscribe pattern with some nice friendly
examples and all that. Putting it in a different document will simply
make it harder to find, IMHO.

/psa

[1] http://tinyurl.com/8pze7




More information about the Standards mailing list