Re: call for agenda items for this weeks telecon (terminology)

On Thu, 2003-02-27 at 19:07, pat hayes wrote:
> >This is the usual weekly call for items for this weeks telecon.
> 
> Regrets for tomorrow. I have to host a meeting here starting at that time.
> 
> >Items on my probables list include:
> >
> >- status of capturing last call comments
> >- scheduling processing them
> >- processing some of them
> >
> >The last of these requires that we have some proposed dispositions, 
> >so I'm looking for suggestions.  I suggest we try to pick some 
> >'easy' (is anything ever easy?) ones so we can start to work up a 
> >rhythm.
> 
> One easy one is getting some nomenclature consistent.
> 
> 1.  What is correct: uriref, URIref, URI Reference ?

RDF URI reference

(blech)

>  Which document 
> or source defines it?


[[[
A URI reference within an RDF graph (an RDF URI reference) is a Unicode
string [UNICODE] that:

      * is in Normal Form C [NFC] and
      * would produce a valid URI character sequence (per RFC2396 [URI],
        sections 2.1) when subjected to the encoding described below. 
]]]
 -- http://www.w3.org/TR/rdf-concepts/#dfn-URI-reference

It seems to go by a number of synonyms...

"... URIs with optional fragment identifiers
(URI references, or URIrefs) ..."
 -- 3.2 URI-based Vocabulary and Node Identification

... but the definition above is what the WG agreed
26th April 2002 regarding
  http://www.w3.org/2000/03/rdf-tracking/#rdf-charmod-uris


Bummer... the syntax spec seems to import using a different
link:
  http://www.w3.org/TR/rdf-concepts/#section-Graph-URIref
from
http://www.w3.org/TR/2002/WD-rdf-syntax-grammar-20021108/#section-Identifiers
ah.. #section-Graph-URIref is the section that dominates
#dfn-URI-reference. So that's clearly the same thing,
technically. Seems worth fixing for editorial
consistency.


Now for a bit of elaboration that goes beyond what you
asked for; perhaps fodder for the issues list, if Brian
or DanBri is so inclined...

The WG has chosen a concept/design which isn't yet available
from mature/ratified specs.

It's not exactly RFC2396's URI reference; that notion
includes "../foo" but that's not a name that RDF interpretations
may give a denotation.

It's not exactly RFC2396's absolute URI; that
doesn't include http://example/foo#bar , which RDF
interpretations *may* give a denotation to.

It's not exactly RFC2396's absolute URI plus optional fragment;
that doesn't include http://example/Andrȷ ,
which may be in the vocab of an RDF interpretation.
This non-ascii stuff is the bit that's too new to
import from any ratified spec.
cf. TAG issue IRIEverywhere-27, cited from concepts section
6.4 RDF URI References).

While we weren't maximally conservative on this issue,
I think we made the right decision. I have already
seen non-trivial deployment[*] of non-ascii characters
in RDF URI references; i.e. I think our decision
minimizes cost over the long term.


* "UTF-8 is needed for this, but that is fine according to
  relevant standards (eikeon: thanks for the pointer) and may
  even help test out some implementations here and there..."
  -- http://dk.space.frot.org/doc/#T2003-02-24


Editorially, I wish we hadn't settled on something with
"reference" in it. I use 'symbol' in my code. Dunno
if that's really best... maybe 'RDF identifier'?
As in 'resource description framework identifier';
almost exactly the same as 'uniform resouce identifier'
except that the latter is us-ascii only to date.

Maybe not sufficiently better to bother... but
here's why I don't like "RDF URI reference"

TimBL has suggested that "URI reference" is
one of the syntactic short-hands for a URI, e.g. "../foo". i.e.
it's a handy piece of local syntax, not the global identifier.

  Rationalizing the term URI Tim Berners-Lee (Thu, Jan 23 2003)
  http://lists.w3.org/Archives/Public/uri/2003Jan/0005.html



Hmm... I don't see the Andre' test cited from the syntax
spec. But it's in the test collection somewhere, no?
Hmm... I can't find it. Jeremy? Jan? Where did
the test cases for rdf-charmod-uris go?

Am I imagining them? no... I remeber this...

  Jeremy to the XML namespace folks...14 May 2002
  http://lists.w3.org/Archives/Public/xml-names-editor/2002May/0003.html

and it leads me to the Apr 26 2002 approval of the tests...
but I don't see them in the test spec.


> 2.   Please can someone provide terminology for 'a well-formed XML 
> literal' (one that covers the cases with and without lang tags) and 
> also for whatever the value space of rdf:XMLLiteral is supposed to 
> contain.

seems to be 'XML document corresponding to a pair ( str, lang )'
from section 5. XML Content within an RDF Graph.
hmm... no hypertext anchor. It's only marked up
with <em> and <var>.



> 3.  Please can we agree that an RDF graph is a set of triples and 
> decide which document is going to give the definition others can link 
> to.

Yes; that would resolve danc-01 nicely.
http://www.w3.org/2001/sw/RDFCore/20030123-issues/#danc-01

Concepts is already that way...

  "An RDF graph is a set of RDF triples."
  -- 6.2 RDF Graph



> Pat
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Thursday, 27 February 2003 23:04:31 UTC