W3C home > Mailing lists > Public > w3c-ietf-xmldsig@w3.org > October to December 1999

Re: Latest draft and Schema / DTD

From: Winchel 'Todd' Vincent, III <Winchel@mindspring.com>
Date: Fri, 12 Nov 1999 14:49:18 -0500
Message-ID: <00f401bf2d47$01f514b0$0a01a8c0@pabdec300ntw>
To: "Joseph M. Reagle Jr." <reagle@w3.org>
Cc: "Donald E. Eastlake 3rd" <dee3@torque.pothole.com>, "IETF/W3C XML-DSig WG" <w3c-ietf-xmldsig@w3.org>, "leg-xml-l" <leg-xml-l@gsulaw.gsu.edu>, <timbl@w3.org>
> That xml.com reference cites things that DTDs don't allow you to do, that
> (at least) would like to do. That includes validating XML namespaces and
> using  mixed/open content models.

Please know, I would like to do this too.  Indeed, it is imperative.

I've been asking a lot of questions on how to do this and no one seems to be
able to give me a good answer.  Some people say that you can do it (validate
more then one XML namespace in the same XML document) with XML and DTDs but
that the tools to do it are not there yet (they are being developed).

What you are saying is that the W3C created XML 1.0 and Namespaces, but they
don't really work.  We led people to believe that would work, but they don't
really, or, at least, not as well as we'd like.  So, now, we've created
XML-Schemas.  XML-Schemas work much better.  Of course, you aren't saying
very loudly that XML-Schemas is a radical break from the SGML world but
you're not denying it either.

All of this is fine and good.  Indeed, for the sake of argument, I'll
concede that XML-Schemas are the best thing since sliced bread and that
XML-Schemas will solve all the worlds problems . . . I note that this is
what the  W3C and the XML community said about XML 1.0 two years ago.

The problem is that you still haven't addressed the larger issue and that is
in regard to the message the W3C is sending to the public about what
technology to learn, use and build on -- now and in the future.  You haven't
given me any confidence that in two months or two years you're not going to
exercise your brilliance (and I do *not* mean this sarcastically) to come up
with another technology that is even better than XML-Schemas.  Nor have you
addressed the issue of simplicity -- HTML, I repeat, has had mass-market
appeal because a 10 year old and a 50 year old can use it without dedicating
their entire life to learn it and build (pay for) tools to support it.
Please understand, I am trying to create technical standards for legal
documents.  What this means is that I must not only be a subject matter
expert in all of the things the W3C does (if I am to use XML), but I must
also be a subject matter expert in the law.  I'd also like to have a free
weekend now and then.

I was in the S/MIME WG at the IETF meeting yesterday and Denis Pinkas
presented an idea about creating machinable policy statements in a way that
is very similar to work I'm doing to automate the policy generation and
mapping of Certificate Policies and Certificate Practice Statements.  Denis
is advocating the use of ASN.1 to code policy.  One person suggested using
XML (a dirty word in the IETF) and this idea was rejected.  One of the
reasons given was that XML was "immature" . . . "perhpas in a year or two
using XML would be possible, but not now."  As an XML advocate, you can
imagine what was running through my mind . . . but now, you are providing
good evidence that, indeed, XML and the specifications surrounding it are,
indeed, immature.

>We (xmldsig) will certainly be mixing
> schemas (embedded signatures and the like).

I agree we will be mixing namespaces.

>  >Now, if I understand correctly, you are saying that using XML DTDs to
>  >validate XML documents is the wrong way to go (contrary to the message
>  >the last two years) and now I (and everyone else) have to switch to XML
>  >schemas and use the phantom tools that exists to make it happen.
> Over time people will begin to use schemas. I'm not telling you have to do
> anything. I am saying TimBL is encouraging WGs to use schemas and to press
> on each other specifications.

So, you seem to be confirming that the message being sent to (received by)
the public is different than the internal W3C agenda.


>  >I assume you will produce examples, so let me add, what about those
>  >who like the idea of SGML and valid documents?  Does this mean the W3C
>  >abandoning valid XML?
> Of course not, people are advancing schemas such that you can have valid
> applications that use XML facilities (like namespaces.)

Are you saying that XML schemas are the only way to use namespaces?   That
is, the use of namespaces is impossible if you want to validate XML from two
(or more) different namespaces in the same XML document against DTDs
corresponding to those namespaces?

If this is what you are saying, then I agree, schemas are the way to go.
But, in this case, I think the W3C has mislead the general public.

>  >Please recall, SGML and valid SGML documents based on
>  >DTDs have been around for 20 years.  As I understood it, the idea of XML
> 1.0
>  >was to simplify SGML to bring the SGML philosophy to the massses.  No
>  >ever said the idea of XML 1.0 was to throw away 20 years of SGML
> philosophy,
>  >experience, and software and replace it with XML schemas.
> Todd, frankly I think you've gone off the handle.

Ad hominem rhetoric is not necessary.  If I am in need of an education, then
please give it to me and I'll back off and be a better person for it.

I note that Donald's comments are similar to mine.  Are you contending that
we are both out of line?

> DTDs were advanced by the
> typesetting community (not theoretical SGML wonks) to ensure people did
> submit documents that were broken and result in dumped jobs. XML
> a lot of the SGML (80% functionality/ 20% complexity) and adds a few new
> features like namespaces. It turns out that in order to support some of
> those XML features your DTDs become extremely complex and still not very
> expressive. That is why folks are working on XML Schemas.

My comments above apply.

I don't recall reading in the W3C Namespace spec that using XML 1.0 DTDs and
namespaces is extremely complex and not very expressive.  Nor do I recall
reading that using XML 1.0 DTDs and W3C Namespaces together was discouraged
in favor of XML-Schemas.

All I want is some clear direction and long-term stability.

What I am contending is not that XML-Schemas are a bad thing, but that you
are radically changing your tune and that by changing your tune now you are
not making me confident that you won't change again in the future.  So, I am
left with the question, why should I use an unstable technology (unstable
suite of technologies) to create legal standards that I would like to have
some longevity?  Very simply, I can't expect every lawyer, judge, and court
administrator to keep up with the W3C's flavor of the month and try to
develop stable legal standards (with the flavor of the month) at the same

Granted I am playing devils advocate to some extent, I note that you did not
respond to these issues in your last reply and I believe it deserves a

Again, tell us what to do and we'll do it, but please provide clear
direction and long-term stability.

>  >> I'd ask that we keep the schema declarations I've used in the draft we
> are
>  >> working on until we post it early next week and the WG can discuss it.
>  >
>  >Is the use of schemas mandated in the XML-Signatures "Requirements"
> document
>  >and/or is the "Requirements" document going to be amended?  Is this up
> a
>  >vote or is it simply going to be done?  Does the W3C consensus policy
> apply?
> Of course, all I've asked is that people have a look at what we've done,
> we can discuss it.
>  >During the second to last telephone conference, you (Joseph) *suggested*
>  >that we try to put the XML-Signature syntax into an XML schema to see
>  >it would look like and asked for volunteers.  But, no one said or ever
>  >suggested we were replacing DTDs with schemas.
> If you use schemas the idea is -- of course -- to get away from DTDs. I'd
> like to post a draft of what the spec likes when using schemas, if the WG
> doesn't like it, we can use DTD declarations.

So, considering you are a co-chair, this means that the XML-Signatures
Requirements document is now unstable?  Indeed, the core signature syntax is
potentially unstable?

Received on Friday, 12 November 1999 14:53:48 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:21:32 UTC