[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: patterns vs. identifiers

• To: xml-dev@l...
• Subject: Re: patterns vs. identifiers
• From: Arjun Ray <aray@n...>
• Date: Mon, 26 Aug 2002 03:32:20 +0000
• In-reply-to: <003001c24cad$6ad9d870$fe193044@tbp1>
• References: <paul@p...> "of Tue, 20 Aug 2002 08:30:47 PDT." <3D6260A7.AB12C5D9@p...> <E17j9Oc-0001lf-00@m...> <no5jmu05nifcftft3src6mlg1mr3flvd0c@4...> <003001c24cad$6ad9d870$fe193044@tbp1>

"Thomas B. Passin" <tpassin@c...> wrote:
 [Arjun Ray]

|| I can't speak for Paul, but I doubt that is what he meant.  I'd say a
|| closer example would be the way in which ID attributes establish identity
|| in a document (i.e. "distinct existence").
| 
| I can't go along with that notion, Arjun. 

I suppose "establish" was the wrong word.  The basic notion would be one
of distinguishability being essential to "identity".

| I say that markup documents establish existence and identity by the 
| existence and context of elements.

I wouldn't go that far.  I view markup as instrumental, not ontological. 

| An ID can make for more robust references (in case of editorial changes) 
| or more convenient ones or more efficient code, but it is not the essence 
| of the thing.

Yes, of course.  But in SGML/XML, the only way to refer to "the same exact
thing" *by authorial intent* is by IDREF attributes (that's what they're
for, after all.)  IOW, suppose you anticipate, on general principles, that
something may need to be refered to (in the sense of "identification")
more than once.  The correct way to use markup, IMHO, is to encapsulate
the knowledge of the referent in an element.  This allows you to give it
an ID attribute, and then reference it when and only when you want or need
to, without possibility of misunderstanding.  Consider

 <stuff>
  <uri id='foo'>http://earth.cosmos.uni/something/i/call/blort</uri>
  <uri id='bar'>http://earth.cosmos.uni/something/i/call/blort</uri>
  <mumble about='foo'>blah blah blah</mumble>
  <blurt about='bar'>yada yada yada</blurt>
 </stuff>

where the 'about' attribute is referential.  I would claim that <mumble>
and <blurt> are about different things as far as the markup is concerned.
String equality of values is, per se, only coincidental.  Compare with

  <stuff>
   <mumble about='http://earth.cosmos.uni/something/i/call/blort'>blah
      blah blah</mumble>
   <blurt about='http://earth.cosmos.uni/something/i/call/blort'>yada
      yada yada</blurt>
  </stuff>

And now you have a *manufactured* problem of whether the 'about' attribute
in each case references the "same thing" or not.  This is B.A.D. (Broken
As Designed).
 
| It is different with RDF, where there are roles for both nodes identified 
| by an identifier and nodes identified by their context (anonymous nodes, I
| mean).

Frankly, I've never understood that issue.  I tried to get it explained
once[1], but I was assured that the Model Theory would take care of it,
and that I had it all wrong anyway.

[1]
http://lists.w3.org/Archives/Public/www-rdf-comments/2001OctDec/0027.html

• References:
  • Re: patterns vs. identifiers
    • From: Paul Prescod <paul@p...>
  • Re: patterns vs. identifiers
    • From: Uche Ogbuji <uche.ogbuji@f...>
  • Re: patterns vs. identifiers
    • From: Arjun Ray <aray@n...>
  • Re: patterns vs. identifiers
    • From: "Thomas B. Passin" <tpassin@c...>

• Prev by Date: Re: Architectural Forms revival?
• Next by Date: Re: patterns vs. identifiers
• Previous by thread: Re: patterns vs. identifiers
• Next by thread: Re: patterns vs. identifiers
• Index(es):
  • Date
  • Thread