headers
MURATA Makoto | 10 Jun 2005 01:04
Picon

Re: The role of media types for XML content

Mark,

(I do think that this topic belongs to the ietf-xml-mime ML.  I am 
ccing to the ietf-types but will not do so any more.)

I think that you have presented some requirements which are not
addressed by any of the existing mechanisms. 

> But I believe the security problem is just one manifestation of the
> layering problem inherrent in using intrinsic dispatch (of which XML
> namespace dispatch in one kind); that an extrinsic dispatch mechanism,
> like media types, cleanly separates the "what" - the raw data - from
> the "how" - the semantics with which that data is to be interpreted.

I do not think that media types are good enough for multi-namespace 
documents and that W3C should develop something different.  Although 
your example can be addressed by specialized media types, my example 
(schemas embedded within XHTML are referenced by XPointer) cannot.

Cheers,

--

-- 
MURATA Makoto <murata <at> hokkaido.email.ne.jp>
Permalink | Reply | 0 comments |
headers
Mark Baker | 10 Jun 2005 16:55
Picon
Favicon
Gravatar

Re: The role of media types for XML content


Hey,

On Fri, Jun 10, 2005 at 08:04:28AM +0900, MURATA Makoto wrote:
> Mark,
> 
> (I do think that this topic belongs to the ietf-xml-mime ML.  I am 
> ccing to the ietf-types but will not do so any more.)

Sounds good.

For those on ietf-xml-mime and not on ietf-types, my original
message can be found here;

http://eikenes.alvestrand.no/pipermail/ietf-types/2005-June/000817.html

> I think that you have presented some requirements which are not
> addressed by any of the existing mechanisms. 
> 
> > But I believe the security problem is just one manifestation of the
> > layering problem inherrent in using intrinsic dispatch (of which XML
> > namespace dispatch in one kind); that an extrinsic dispatch mechanism,
> > like media types, cleanly separates the "what" - the raw data - from
> > the "how" - the semantics with which that data is to be interpreted.
> 
> I do not think that media types are good enough for multi-namespace 
> documents and that W3C should develop something different.

I think media types are fine for multi-namespace documents.  I just
think that XML, by itself, doesn't provide a sufficiently rich framework
(Continue reading)

Permalink | Reply | 17 comments |
headers
MURATA Makoto | 11 Jun 2005 08:50
Picon

Re: The role of media types for XML content


Mark,

Mark> >Although 
Mark> > your example can be addressed by specialized media types, my example 
Mark> > (schemas embedded within XHTML are referenced by XPointer) cannot.
Mark> 
Mark> Can you elaborate on that example please?  I don't think I've seen any
Mark> schemas embedded in XHTML before.

Consider the following XML document (say multi.xml).  This example is not 
artficial.  It is useful for literate programming.  As you know better
than me, W3C is trying to promote compound documents.

I would like to reference to the embedded RNG schema using multi.xml#rng. 
What is the media type of this document?  And do you want to specify a 
specialized media type when you reference the embedded RNG?

<?xml version="1.0" encoding="UTF-8"?>
<html xmlns="http://www.w3.org/1999/xhtml">
    <head xml:lang="en">
        <script type="JavaScript"/>
        <title dir="ltr" lang="en"/>
        <style type="das"/>
    </head>
    <body>
        <h1>Welcome to XHTML</h1>
        <div title="a RELAX NG grammar">
            <grammar 
                xml:id="rng"
(Continue reading)

Permalink | Reply | 16 comments |
headers
Mark Baker | 13 Jun 2005 17:25
Picon
Favicon
Gravatar

Re: The role of media types for XML content


Hey,

On Sat, Jun 11, 2005 at 03:50:03PM +0900, MURATA Makoto wrote:
> Mark> >Although 
> Mark> > your example can be addressed by specialized media types, my example 
> Mark> > (schemas embedded within XHTML are referenced by XPointer) cannot.
> Mark> 
> Mark> Can you elaborate on that example please?  I don't think I've seen any
> Mark> schemas embedded in XHTML before.
> 
> Consider the following XML document (say multi.xml).  This example is not 
> artficial.  It is useful for literate programming.  As you know better
> than me, W3C is trying to promote compound documents.

Right.  For those who weren't aware, the W3C has a Compound Document
Formats WG[1] that is trying to address some of these issues.  I'm on
the WG as a representative of Justsystem Corp.

> I would like to reference to the embedded RNG schema using multi.xml#rng. 
> What is the media type of this document?

You could use the text/html or application/xhtml+xml media types.  If
fallback rendering was a problem, you could add some styling directives,
or else just "display: None"-it and place an HTML description in there.
Fragment identifiers would work as expected.

You could use application/xml too, though as you know, you don't get
fragment identifers.  Also, IE doesn't make the same assumptions as
other browsers[2], and will only display the tree view.  Moreover, the
(Continue reading)

Permalink | Reply | 15 comments |
headers
MURATA Makoto | 14 Jun 2005 16:47
Picon

Re: The role of media types for XML content


Mark> >And do you want to specify a 
Mark> > specialized media type when you reference the embedded RNG?
Mark> 
Mark> Ideally, yes, I think that would be valuable for the reasons I gave
Mark> before concerning layering and security.
Mark> 
Mark>  [1] http://www.w3.org/2004/CDF/
Mark>  [2] http://www.markbaker.ca/2004/01/XmlDispatchTest/

You propose to use a specialized media type for a fragment and 
to use a different media type for the entire resource.  It is a very 
interesting idea.  It might help the marriage between media types and 
multi-namespace XML documents.  But I do not think that MIME RFCs or
URI/IRI RFCs bless it (at least now).  Media types have been intended 
as values of the content-type field of MIME.  

You might want to raise this issue in the W3C CDF WG or W3C TAG, and 
you might even want to write an RFC that updates MIME RFCs or URI/IRI
RFCs.

However, unless such significant changes are endorsed, I continue 
to be very reluctant to register a specialized media type for 
the RELAX NG XML syntax.

After all, it is ISO/IEC JTC1/SC34 that is requesting the registartion
of a media type for the RELAX NG compact syntax.  In its last meeting, 
SC34 decided to request this media type, but did not consider a specialized 
media type for the RELAX NG XML syntax.  Since we are talking about 
media types in the standard tree, the media type for the RNG compact
(Continue reading)

Permalink | Reply | 14 comments |
headers
Chris Lilley | 14 Jun 2005 17:40
Picon
Favicon

Re: The role of media types for XML content


On Tuesday, June 14, 2005, 4:47:40 PM, MURATA wrote:

Mark>> >And do you want to specify a 
Mark>> > specialized media type when you reference the embedded RNG?
Mark>> 
Mark>> Ideally, yes, I think that would be valuable for the reasons I gave
Mark>> before concerning layering and security.
Mark>> 
Mark>>  [1] http://www.w3.org/2004/CDF/
Mark>>  [2] http://www.markbaker.ca/2004/01/XmlDispatchTest/

MM> You propose to use a specialized media type for a fragment and 
MM> to use a different media type for the entire resource.  It is a very
MM> interesting idea.  It might help the marriage between media types and
MM> multi-namespace XML documents.  But I do not think that MIME RFCs or
MM> URI/IRI RFCs bless it (at least now).  Media types have been intended
MM> as values of the content-type field of MIME.  

MM> You might want to raise this issue in the W3C CDF WG or W3C TAG, and
MM> you might even want to write an RFC that updates MIME RFCs or URI/IRI
MM> RFCs.

his is interesting, and speculative, future work. However, and not
distracting from the original specific issue:

MM> However, unless such significant changes are endorsed, I continue 
MM> to be very reluctant to register a specialized media type for 
MM> the RELAX NG XML syntax.
(Continue reading)

Permalink | Reply | 8 comments |
headers
MURATA Makoto | 15 Jun 2005 01:06
Picon

Re: The role of media types for XML content


Chris> 
Chris> MM> However, unless such significant changes are endorsed, I continue 
Chris> MM> to be very reluctant to register a specialized media type for 
Chris> MM> the RELAX NG XML syntax.
Chris> 
Chris> I don't see how the issue of how/whether to point to a portion of an xml
Chris> resource which happens to be in RNG, has *any* bearing on the issue of
Chris> what media type to use when RNG is served as a stand-alone document,
Chris> which is by far the typical case at the moment. There is no dependency
Chris> there.

I have said that specialized media types for XML-based vocabularies are 
ad-hoc solutions which do not work for multi-namespace documents and
that proliferation of specialized media types blocks future use 
of multi-namespace documents.  I thus see strong dependency.

Chris> MM> After all, it is ISO/IEC JTC1/SC34 that is requesting the
Chris> MM> registartion of a media type for the RELAX NG compact syntax. In its
Chris> MM> last meeting, SC34 decided to request this media type, but did not
Chris> MM> consider a specialized media type for the RELAX NG XML syntax.
Chris> 
Chris> 
Chris> So, now I am asking them to consider it. Will you convey this request to
Chris> them, or do I need to convey it by a more formal mechanism?

I can formally raise this issue in the next SC34 meeting in November and
can also informally contact SC34 members now.  However, between now and
Novembmer, SC34 WG1 can do nothing.  ISO/IEC JTC1 has no mechanisms for
making decisions other than formal meetings.
(Continue reading)

Permalink | Reply | 7 comments |
headers
Liam Quin | 15 Jun 2005 01:59
Picon
Favicon

Re: The role of media types for XML content


On Wed, Jun 15, 2005 at 08:06:36AM +0900, MURATA Makoto wrote:
[...]
> I have said that specialized media types for XML-based vocabularies are 
> ad-hoc solutions which do not work for multi-namespace documents
Here I agree with you.  The architectures clearly don't match at all:
Internet media types simply don't work for this sort of document today.
Mozilla Firefox is making this clear -- it supports XHTML with
embedded MathML and SVG for example, in either order, and perhaps
with XHTML and SVG and mathML inside the SVG...

> and that proliferation of specialized media types blocks future use 
> of multi-namespace documents.
I don't think it's blocking future use.  I think in the long term we
have to rethink how multi-namespace XML documents are handled.

But that should not stop us from using the existing infrastructure as
best and as helpfully as we can in the meantime.

I'd like to see a registration for the XML syntax simply because
it's useful today, despite the limitations.

Best,

Liam

--

-- 
Liam Quin, W3C XML Activity Lead, http://www.w3.org/People/Quin/
http://www.holoweb.net/~liam/
(Continue reading)

Permalink | Reply | 6 comments |
headers
MURATA Makoto | 15 Jun 2005 02:45
Picon

Re: The role of media types for XML content


Liam> I'd like to see a registration for the XML syntax simply because
Liam> it's useful today, despite the limitations.

Then, why doesn't W3C register specialized media types for XSLT or 
W3C XML Schema?  Is there some consensus in W3C?

Anyway, we can do nothing about a specialized media type for the RNG 
XML syntax by November.

Cheers,
--

-- 
MURATA Makoto <murata <at> hokkaido.email.ne.jp>
Permalink | Reply | 3 comments |
headers
Martin Duerst | 15 Jun 2005 02:32
Picon
Gravatar

Re: The role of media types for XML content


At 08:59 05/06/15, Liam Quin wrote:

 >But that should not stop us from using the existing infrastructure as
 >best and as helpfully as we can in the meantime.
 >
 >I'd like to see a registration for the XML syntax simply because
 >it's useful today, despite the limitations.

I fully agree with Liam. I think the argument against registering
a type for the XML syntax was based on the question "but what about
pieces of this in another XML document?". Creating a type for a
standalone Relax NG document in XML syntax, which allows people
to label such documents if they can, does not exclude the use
of Relax NG in XML syntax as part of other documents.

Also, Relax NG in XML syntax most probably may have provisions
of some sort about how to include other namespaces, e.g. for
documentation,...

Regards,    Martin.
Permalink | Reply | 1 comment |

Gmane