W3C home > Mailing lists > Public > www-rdf-logic@w3.org > July 2002

Re: Input sought on datatyping tradeoff

From: Drew McDermott <drew.mcdermott@yale.edu>
Date: Sat, 13 Jul 2002 11:53:25 -0400 (EDT)
Message-Id: <200207131553.g6DFrPI29540@pantheon-po02.its.yale.edu>
To: www-rdf-logic@w3.org


   [me]
   >We've been assuming that for
   >every property (here, dc:property in particular) there is a parser
   >that unambiguously specifies how to interpret the strings that occur
   >as its values, even if we don't know what that parser is.

   [Brian McBride]
   I wasn't aware I was making that assumption, but maybe I was.  I think I 
   was assuming that the model theory will say there must be something in the 
   domain of discourse the _:l binds to.

I think the assumption is inevitable if Idiom 2 is going to stick
around, which it apparently is.  The alternative is to say that RDF
triples *always* have an uninterpreted string as their third element,
which means that Idiom 1 is no longer correct.  Since Idiom 1 is
obviously better than Idiom 2, that would be a big mistake.  

By the way, let me clarify what I am saying so that it doesn't come
off as implying that XML must be junked, which I acknowledge is out of
the question.  I am perfectly content to allow notations such as

   <rdf:Description...>
      <abc:ageInYears>10</abc:ageInYears>
   </rdf:Description>

just as now.  It's just that 10 would be parsed as a decimal integer.
(Binary integers would have some other notation, such as "10b2" .)  If
you wanted a string, you would put quotes in:

   <rdf:Description...>
      <abc:title>"10"</abc:title>
   </rdf:Description>

Alas, we have to confront the fact that properties can be moved into the
attribute list, when quotes reappear.  We could use single quotes for
that purpose:

      <abc:film .... abc:ageInYears='10' abc:title='"10"' />

For N3, the issue doesn't come up, so we always dispense with
superfluous quotes:

     <abc:film> <abc:ageInYears> 10 ; <abc:title> "10" .

We still have to have conventions to handle strings containing awkward
characters such as angle brackets and quotes.  Here I don't know what
to propose, because the SGML entity notation seems to be required, and
it makes no sense to me.  (In XSLT, people are apparently willing to
write "less than or equal" as "&lt;=" instead of "<=".  One would
think the quotes would suffice to indicate that the "<" is not part of
a tag, but the lexical conventions of SGML apparently trump the
quotes, which come in only at the XML layer.  How is it that in, say,
the Fortran world people have found a way for the language to evolve
gracefully, and in the SGML world evolution is viewed as impossible?)

                                             -- Drew McDermott
Received on Saturday, 13 July 2002 11:53:30 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:42 GMT