SV: [codelist] Fwd: [chairs] FW: New Specification Templates and Guidelines now Available

["Bryan Rasmussen" <BRS@itst.dk>]

Hi again,

I suppose the part about rejecting the W3C dogma that everything must be
dereferencable would indicate an objection (although unexplained) to the Sean
McGrath comment below. 

At any rate one must have a rule to determine if two parties are talking
about the same thing. 
For XML Namespaces the rule is basically that the strings are the same. This
is not changed by third party specifications adding dereferencing
capabilities on top of namespaces in general or any namespace in specific.

As noted your info scheme seems to have added some rules on top of the XML
Namespaces rules for determining if one is talking about the same thing,
specifically ways to look up the meaning of an info uri via a registry so
that someone putting info:schemainfo inside of their xml can, by reference to
the registry, be shown that what they mean by info:schemainfo was not what
everyone else meant by info:schemainfo and they should change it (note: have
not checked but assume there is not actually a info:schemainfo uri registered
in your registry)


Cheers,
Bryan Rasmussen
-----Oprindelig meddelelse-----
Fra: Bryan Rasmussen [mailto:BRS@itst.dk]
Sendt: 27. februar 2007 09:30
Til: Ray Denenberg, Library of Congress; Code List Representation TC
Emne: SV: [codelist] Fwd: [chairs] FW: New Specification Templates and
Guidelines now Available


Well, that something is dereferenceable does not obligate it to be
dereferenced. It is true that the common problem when people are first
introduced to a namespace beginning with http is that they expect it to be
required to be dereferenced for the namespace to function, and sometimes it
can be difficult to get it through to them that it is not so, but sooner or
later most people get it, so I don't think one can say namespaces should
never be http. Furthermore since the namespaces spec allows http I think it
should have been understood that while they did not need to be dereferenced,
implied dereferenceability was an attribute of any namespace uri using http
as the scheme identifier. If one has made a http uri, which seems to be the
most common thing people do, it will at any rate return some information if
dereferencing is attempted, so why not have it return something meaningful?

Finally, I can make an implementation quite easily of the info uri scheme on
windows that maps to
http://info-uri.info/registry/OAIHandler?verb=GetRecord&metadataPrefix=reg&id
entifier={ the input info uri }
such that info:arxiv 'dereferences' to
http://info-uri.info/registry/OAIHandler?verb=GetRecord&metadataPrefix=reg&id
entifier=info:arxiv/

so much for purity. 

And why does one have a registry for info: identifiers? because one wants to
find out the use of the namespace, I guess. This reminds me of a comment by
Sean McGrath
http://seanmcgrath.blogspot.com/2005_04_10_seanmcgrath_archive.html#111337910
941603816#111337910941603816

"Any would-be rigid designator, needs a state of affairs to be dereferenced
against. HTTP is a well known resolver of rigid designators given a state of
affairs:-)

Most ironic in all the MMTT URN shenanigans is that so many URN resolving
schemes end up doing a quick shimmy to convert the URNs into unique HTTP URIs
:-)"
(what I considered most relevant of that comment)

I would argue that what you are doing is basically limiting the functionality
of something and asserting it is better. The assertion that a less powerful
thing is an improvement is claim requiring some proof, and given the last few
years it does not seem anyone has a good proof. I suppose this is because the
people arguing for dereferenceability can point to things that can be done if
one dereferences, and the ones against dereferenceability cannot point to
anything they can do that the dereferencing people can't do. This, imho, is a
bad position to be in when making an argument. 


Cheers,
Bryan Rasmussen




-----Oprindelig meddelelse-----
Fra: Ray Denenberg, Library of Congress [mailto:rden@loc.gov]
Sendt: 26. februar 2007 20:39
Til: Code List Representation TC
Emne: Re: [codelist] Fwd: [chairs] FW: New Specification Templates and
Guidelines now Available


Don't mean to spoil the fun but please put me down in the NO column.

I intensely dislike the use of http: URIs for xml namespace names - for that
matter, for  "pure" identifiers in general - identifiers who's sole purpose
is to allow two parties to  agree that they are talking about the same
thing.   And I reject the w3c dogma that all identifiers must resolve to
something, that there is no such thing as a "pure identifier".

XML namespace names should not be dereferenceable, and thus should not be
http:

In fact, we (Library of Congress) have applied for an 'info:' namespace, see
http://info-uri.info/registry/OAIHandler?verb=ListRecords&metadataPrefix=oai_
dc

for xml namespace names.  Pending approval of that you could coin a URI like
this, say, for the codelist schema:

                          'info:xmlns/oasis/codelist'

Simple as that.

You can see the list of pending 'info:' namespaces at :

http://alcme.oclc.org/wikid/CollectionInfoUriRegistry

and in particular the xmlns application at:

http://alcme.oclc.org/wikid/CollectionInfoUriRegistry:info:xmlns/

--Ray

----- Original Message ----- 
From: "Bryan Rasmussen" <BRS@itst.dk>
To: <abcoates@mileywatts.com>; "G. Ken Holman"
<gkholman@cranesoftwrights.com>; "Code List Representation TC"
<codelist@lists.oasis-open.org>
Sent: Monday, February 26, 2007 12:41 PM
Subject: SV: [codelist] Fwd: [chairs] FW: New Specification Templates and
Guidelines now Available


> I would give a +1 to that as well.
>
> crossing my fingers that oasis acceptance means imminent Danish acceptance
of
> RDDL (by imminent I mean after months of meetings but nonetheless)
>
> Cheers,
> Bryan Rasmussen
>
> -----Oprindelig meddelelse-----
> Fra: Anthony B. Coates (Miley Watts) [mailto:abcoates@mileywatts.com]
> Sendt: 26. februar 2007 18:40
> Til: G. Ken Holman; Code List Representation TC
> Emne: Re: [codelist] Fwd: [chairs] FW: New Specification Templates and
> Guidelines now Available
>
>
> I'm quite happy for us to do that.  Cheers, Tony.
>
> On Mon, 26 Feb 2007 15:45:46 -0000, G. Ken Holman
> <gkholman@CraneSoftwrights.com> wrote:
>
> > Just a quick note regarding the new OASIS guidelines for documentation.
> >
> > Note that URN protocol strings are still allowed for namespace URI
> > strings, but with the adoption of RDDL attributes does it make sense for
> > us to move to an HTTP protocol string for the code list URI string so
> > that we have RDDL documentation at the URL of the URI?
> >
> > (try to say that three times fast!)
> >
> > . . . . . . . . . . . . . Ken