xml-dev - Re: Microsoft's DISCO proposal and XML packaging

[Rick JELLIFFE <ricko@geotempo.com>]

Edd Dumbill wrote:
> 
> I've just read http://msdn.microsoft.com/xml/general/disco.asp
> "Draft: Discovery of Web Services (DISCO)"

It looks way better than nothing. Packaging or resource discovery seems
to
be one area where the commitment by vendors to open standards seems
to come up against the realization that they lose a barrier to
competition.
So it is particularly heartening if MS can push something through in 
that area: I think we should welcome any initiative here. 

I would guess that the design of DISCO allows that: it doesn't require
any (co-operative) controlled vocabulary, so there is still complete
scope for each vendor to use different namespaces for the same thing. 

It seems that the format is intended to work mainly by linking: if you
really want to discover what the resource is, you have to fetch it and
poke around.  

> I will also note that DISCO seems close to Sean McGrath's XFM (XML
> Feature Manifest) and Simon St.Laurent's XPDL
> <http://purl.oclc.org/NET/xpdl>.  I asked at WWW9 about W3C
> initiatives for addressing this kind of "packaging" activity, and got
> the response that it was effectively on hold and nothing was happening
> on it

People interested in this might also see my 

 * "DrLove: Document Resource  Locations (on Valid Elements)" 
 at  http://www.ascc.net/~ricko/drlove.htm

which proposes
 1) use a PI like the stylesheet PI locating a "resource dictionary"
 2) the resource dictionary is a simple wrapper element
 3) the things that are wrapped should be RDF statements

So the differences of DISCO with DrLove for these three things are:
 1) Disco has more comprehensive rules which do not involve defining a
new PI, 
kowtowing to fashion;
 2) same, thoughh fewer attributes
 3) the thing that is wrapped is a lot less ambitious, just a link.

I think DISCO's #1 makes sense politically. I think DrLove's #3 (use
RDF) is 
better.  However, I note that DISCO is open, so RDF statements can be
put in 
if needed. 

So my recommendation, for what it is worth, for DISCO would be to
reconcile
it either with XLink or with the RDF.  Decide whether DISCO:discovery
should
contain links or resource descriptions (or, best, both) and use it as a
good
opportunity to support the W3C specs for links and for resource
descriptions.

I wonder if Simon and Sean might care to compare Disco with their
approaches?


Rick Jelliffe