Re: FW: PIDF - using xml schema

["Yaron Y. Goland" <modulus@xxxxxxxxxxx>]

Unfortunately I lack sufficient intelligence to write XML schemas unaided.
I am forced to use a tool like XML spy. On the other hand, I was able to
write a correct Relax NG schema after just quickly skiming the draft using
windows Notepad.  Personally I doubt that there exist any two truly
interoperable implementations of XML schema, at least if you try to use
features beyond the very core set. I strongly dislike XML schema, I think
it is a hinderance to protocol development, to interoperability and to
good design. But unfortunately it may be "good enough".

That having been said a few facts are quite clear.

1. The only widely accepted XML type system is the one provided by XML
schema. If memory serves (and it probably doesn't) even Relax NG points
readers there.

2. XML Schema has enormous support, mostly because the alternatives are
not seen as being credible. Most folks see the W3C as the gospel of XML
standards and are unwilling to adopt anything XML that hasn't been blessed
by it. The end result is that all the major commercial XML houses
(Microsoft, IBM, Sun, etc.) support XML schema as their primary schema
mechanism. Furthermore WS-I, which much to my surprise is not just an
IBM/Microsoft pet forum, will (I predict) adopt XML schema as part of its
basic profile.

Near as I can tell (and much to my disapointment) XML schema seems a done
deal. I would love to be convinced otherwise. However, if indeed XML
schema (as deeply flawed as it is) has won then perhaps the most value we
can add is to provide a guide to which features in it should be used based
on:
	A) How well they are supported across various systems
	B) How likely they are to produce useful schemas.

I have already made a stab at this in my own comments -
http://www.goland.geek/#04032002 or http://www.goland.org/#04032002. BTW,
I fully expect that WS-I will perform this same exercise.

	I wish there was another way, any ideas?

		Yaron

On Mon, 22 Apr 2002, Carr, Wayne wrote:

>
> This didn't get sent to the list because of anti-spam - so here it is again.
>
> -----Original Message-----
> From: Carr, Wayne
> Sent: Monday, April 22, 2002 12:33 PM
> To: 'bateman@xxxxxxx'; LMM@xxxxxxx
> Cc: Carr, Wayne; ietf-xml-use@xxxxxxx
> Subject: RE: PIDF - using xml schema
>
>
> > In terms of difficulty, Wayne might be better placed to give
> > an opinion.  From my standpoint it is both useful and relatively
> straightforward.
>
> Hi Larry,
>
> Writing the xml schema for pidf was trivial.  I wrote it as I read the pidf
> draft for the first time to understand the format.  I think it is also much
> easier for someone not familiar with either dtd's or xml schema to read a
> simple schema than a dtd.  It's a good way to understand precisely what's
> permitted in the format.
>
> As Adrian pointed out, pidf is using an extensibility mechanism like SOAP
> uses - extensibility is a key part of the design - and XML Schema lets you
> specify that precisely.  DTD's don't.  DTD's are also a real pain to be able
> to use properly with namespaces, wheras XML Schema takes namespaces into
> account.  So all around XML Schema is a better way to define something like
> this.
>
> 	Wayne
>
> > -----Original Message-----
> > From: Adrian Bateman [mailto:bateman@xxxxxxx]
> > Sent: Sunday, April 21, 2002 2:40 PM
> > To: LMM@xxxxxxx
> > Cc: 'Carr, Wayne'; ietf-xml-use@xxxxxxx
> > Subject: RE: PIDF - using xml schema
> >
> >
> > On 21 April 2002 17:14, Larry Masinter wrote:
> > >
> > > > I've just finished reading
> > > > draft-hollenbeck-ietf-xml-guidelines-00.txt
> > > > and have the following observations for consideration:
> > >
> > > Version -00 was our first shot, there's now an -01,
> > > and there's significant discussion over the applicability
> > > of XML schema. It would be very helpful to get your review
> > > on the guidelines:
> >
> > Hi Larry,
> >
> > I've scanned through the -01 document now. As far as the discussion of
> > XML Schema goes, I think it all makes sense thus far. My preference is
> > for XML Schema for two reasons, the first a personal
> > deficiency: I don't
> > really understand all the DTD syntax, just about enough to
> > read one, but
> > certainly not to write one. The second is more practical: for PIDF in
> > the IMPP group we have described our extensibility mechanism using
> > namespaces. Clearly DTD's don't handle this and XML Schema seems an
> > ideal tool to address this.
> >
> >
> > >  Which guidelines are helpful? Which don't make sense or
> > > have confusing applicability?
> >
> > I pretty much skipped the sections on XML Selection Considerations and
> > XML Alternatives. I guess they do have a place in this document, but
> > weren't really relevant for the point we've reached in IMPP (and I
> > suppose I already have some of my own views about XML applicability
> > ;^>).
> >
> > Aside from that, I think everything in section 4 is useful
> > and thinking
> > back a couple of years, I think if we'd had this document
> > then, a lot of
> > pointless argument might have been avoided in IMPP (perhaps
> > not though,
> > but at least we'd have had some ground rules, and in this context the
> > sections on XML Selection, etc. would have been useful).
> >
> > > Are there other topics that you wish were covered?
> >
> > One thing that affects us that is mentioned in the text is how to
> > determine extensions that are mandatory to recognise. As you
> > say, there
> > is no standard mechanism in XML, but perhaps an example might be given
> > which would tend to help protocol authors end up with the same or
> > similar solutions which can't be a bad thing?
> >
> > > There's been significant pushback on the recommendation to use XML
> > > Schema (from Relax NG proponents). How useful/difficult was it to
> > > define an XML Schema?
> >
> > I imagine different groups might wish to present different
> > recommendations. The suggestion to treat XML Schema as first
> > preference
> > seems sensible to me, but I haven't really encountered many
> > of the other
> > mechanisms (which can be interpreted in different ways and others can
> > draw their conclusions from that).
> >
> > In terms of difficulty, Wayne might be better placed to give
> > an opinion.
> > From my standpoint it is both useful and relatively straightforward.
> >
> > Best regards,
> >
> > Adrian.
> >
> >
>