Re: XML observation

Hello Graham,

At 16:56 03/07/08 +0100, Graham Klyne wrote:

>At 10:01 08/07/03 -0400, Martin Duerst wrote:
>>>There a couple of other data I'd like to introduce:  in his 
>>>pre-last-call review of the RDF concepts document, Tim Berners-Lee 
>>>expressed some concern about the "distinguished" nature of XML in RDF 
>>>[1].  That was a comment which was never really given a lot of attention 
>>>that I was aware of, and is one which I personally agree.  Similar 
>>>sentiments have been expressed in an exchange between Patrick and Peter 
>>>Patel-Schneider [2].
>>
>>I discussed this with Tim a few weeks ago. He was not aware of
>>the fact that you don't need an infoset to handle XML literals
>>in RDF.
>
>Hmmm... I'm not sure what the implication of this may be.
>
>RDF/XML syntax is (now) defined in terms of the XML infoset
>of the containing XML document.  So, I think, that suggests that
>XML literals in RDF/XML must likewise be defined.  Or maybe not,
>because RDF literals are defined (in some cases partly) in terms
>of the Unicode string that represents the literal.
>At best, it feels rather convoluted.

Sorry for my imprecision. I think I should have said that
Tim wasn't aware of the fact that there was no need to use
the Infoset in an RDF store. An RDF parser will of course
use the infoset in whatever way it is exposed by the API
of the underlying XML parser when it parses RDF/XML.


>>And the inclusion of XML literals in RDF is not a mistake, it's
>>important for i18n reasons. I feel it's terribly sad that this
>>gets ignored here time and again.
>
>You don't quite say this, but this suggests the purpose of XML
>literals in RDF is for I18N.  If so, this would be the first time
>I've heard that suggested.

It is true that I18N was a major advocate for including them
in RDF in the first place. That does not mean that XML literals
in RDF are for I18N only, of course. But it hopefully makes it
understandable that changes to the original design (such as
throwing out xml:lang from XML literals, disconnecting plain
literals and XML literals,...) are seen in a very bad light
on our side.


>This is very suggestive of the role of XML as text-with-markup,

Yes indeed. As I said already, that's what XML literals
were designed for, because there was no other standard way
to include text-with-markup into RDF, and because text-with-markup
was considered too important a thing (expecially for i18n) to
just drop.


>and
>to my mind does little to square that role with XML-for-arbitrary-data.

Yes, it doesn't. And for RDF, it doesn't have to. If you have
structured data, and you want to handle it in RDF, you should
use RDF, not XML. I'm surprised that I should have to tell you.


>I find this lack of clarity about the role of XML to be really unsettling,
>because, as Pat has pointed out, there are some real tensions here.

The tension is here. But RDF has the chance to absorb this
tension in an elegant way, rather than to increase it.


>(Hmmm... general W3C process point:
>given I18N is regarded as so fundamental, I think the purpose would be
>better served by having I18N *participation* from the beginning, or
>at the beginning, not occasional reviews later in the process.)

As Misha already pointed out, I18N was involved from early on
in the M&S days. Misha was a full participant in the RDF WG.
Also, there was a joint meeting at the Technical Plenary in
Cannes in Feb or March 2002.

Regards,    Martin.

Received on Tuesday, 8 July 2003 17:34:05 UTC