RE: Dictionares in the library

At 09:45 AM 2000-05-21 -0400, Mike.Champion@SoftwareAG-USA.com wrote: 

>
> -----Original Message----- 
> From: Tim Berners-Lee 
> To: xml-uri@w3.org 
> Sent: 5/21/00 8:53 AM 
> Subject: Dictionares in the library 
>
> > T: But Achilles, for the last time, a dictionary is NOT  A  BOOK! 
>
> Hmmm.  I thought our discussion here was about much less profound matters.
> Our Tortoise says that books should reference the language they are written
> in by some unique and persistent string, whereas Achilles says they must use
> the ISBN of the dictionary that contains their vocabulary. <grin>


Thank you for putting things in a way over which we can share grins.

Here are some propositions.  How should these plattitudes be implemented
a.k.a.
enforced?

A. If the book uses language that isn't in almost every dictionary, and there
is some dictionary which explains the language in dictionary form, the book
should cite this dictionary.

B. If the dictionary has been assigned an ISBN number, this number should be
disclosed in the citation in the book using the terms for which good
explanations appear in that dictionary.

I really think that is as far as the book analogy goes.  The devil, of course,
is in the details.

I don't expect much objection to these as hypothetical moral plattitudes.
I do
expect some mutual pushing and shoving over implementation/enforcement.
What I
unfortunately hear here is a lot of miscommunication-type friction over
implementation mechanics that isn't really nececessary to the genuine policy
debate.

I don't yet understand, amid all the heated debate, what level of support I
would find for the proposition "it is not the intended use of XML that people
should use or introduce markup names that lack dictionaries."   That is
roughly
the commitment that the WAI is looking for under the guideline "3. Export
semantics" in the rough working draft [but best single document on the
subject]
found at <<http://www.w3.org/WAI/PF/xmlgl>http://www.w3.org/WAI/PF/xmlgl>.

Note:  As a personal opinion, I would have to say that the WAI need does not
get so precise as to say "there must be a recovery key for a dictionary in the
ns-attr."  Not at all.  If the dictionary follows the language established by
the XML Schema drafts and the citation uses the schema location notation also
defined there, so much the better.  The kind of language used in the
dictionary, and the way to cite the dictionary from the book using the
sublanguage [module] need to be flexible.  The architecture needs to support
alternative forms of each of these.

The use of a link to a dictionary as the string in the namespace
declaration is
good or bad depending on the form of documentation that comprises the
dictionary in question.  It should be a possible practice for simple
documentation cases and equally it should not be required, to handle more
complex documentation cases where the pertinent documentation, for example, is
a collection without a published catalog or defined root element.

On the other hand, in my reading the statement "the intended use is that
namespaces be identified by 'some unique and persistent string' and that
anything further in terms of providing or linking to a dictionary is 'entirely
apart from the intended use' is seriously hostile to the accessibility
interest
in this matter.  Namespaces are rarely just a namespace, and the
linking-to-dictionary requirements should be set to support a maximal-depth
processing profile reflecting the actual definition-and-documentation level of
the markup dialect used, not the minimal processing profile that applies
across
all markup dialects using the namespace syntax.

Al

Received on Sunday, 21 May 2000 13:28:46 UTC