glossary: 'Schema' and schemas

At 12:16 PM 2001-04-24 +0100, DPawson@rnib.org.uk wrote:
>Henry Thompson suggests
>
>W3C XML Schema (language).
>
>as the W3C 'version' of schemas.
>
>Could we add that to the terminology?
>

** Summary

Yes, we should have stuff in our lexicon clearly identifying this language.

Yes, Henry's exquisitely crafted phrase provides a form of reference which is
largely swift, transparent, and sure.

But 'adding' that to "the terminology" falls short of what we need to do.  For
the XMLGL document in particular, and for the WAI knowledge base more broadly,
we need to link this phrase into a pattern of usage including forms of
reference that are variously more swift, more transparent, and more sure.
Read
on.

We need multiple forms of identification or reference to this language, all
associated with the same language definition and with one another.  Some
possible forms of reference are offered below.

[Background.  We wish to include the definition of this language to our
resource base that the XMLGL draws on.]

a) "XML Schema"  -- for recurring use in a document which refers to it a lot; 
I will even use just 'Schema' with a capital S where there are only two
choices, the general meaning and the specific meaning indicated [more surely]
by b) and c) following.

b) "<logo>W3C</logo> <docTitle>XML Schema</docTitle> language" -- for
occasional reference in a document that doesn't refer often.

c) A language defined by the W3C for writing schemas which are both in and
about XML, see <<http://www.w3.org/XML/Schema>http://www.w3.org/XML/Schema>.

Either a) or b) as appropriate may be used to refer to the language identified
by c).  Form b) may often be used unsupported by any further explanation.
Form
a) can be used in a context where the reference is endemic; but it should be
backed up by definition along the lines of c) in said context.

Note also that a precise document version has not been cited in the above. 
Many references, and names among them, refer to ambiguity groups or regions. 
This is obvious in the case of lowercase s schemas.  In the case of capital S
Schema it is still true, while less obvious.  It is intended that the terms as
set forth above refer inclusively to various language specifications in
successive drafts as reflecting a common intent to define and eventually use
one language.  We are not distinguishing between different draft versions of
the language definition in the above forms of reference.  Schema instances
which are constructed using early drafts of the language specification are
included in what we mean when we refer to XML Schema.  That is why I used the
URL for the overview page and not an URL referencing a particular dated draft.

Al

** rant (may be skipped)

For our document it is inconceivable to _define_ what we mean by this
reference
without an URI.  While not everything worth referring to will have a URI,
_this
language_ does have a URI _and we should use it_.  We should also not use this
reference without some more explanation, even if it is just a "q.v." attached
to the bibliographic citation.  [Quod vide, which see; this means that the
cited resource should be accessed for required explanation.  This is old
fashioned markup for a normative reference.  It indicates that what is being
said locally may be misunderstood if what is stated remotely has not been
digested first.]

Of course, Henry's suggestion _is_ diagnostic.  If one prunes off the
parenthetical expression, and submits the remaining "W3C XML Schema" to Google
with an "I'm feeling lucky" one is indeed lucky.  One gets to the right
place. 
So that much description is sufficient, for practical purposes.  But for a
document where we actually depend on the correct expansion of this reference,
just putting this phrase in our lexicon is not sufficient.  Something more on
the order of c) above must be included, since we are talking about a language
that has a published definition.  We should make a complete bibliographic
reference to the published definition within the local data of our utterance.

In my email I have stopped providing URLs for many things where Google
suffices
to find the appropriate background.  I don't think we want to go quite that
far
in a formal document.

The whole point of a dictionary is not the symbols nor the definitions but the
association between specific symbols and specific definitions or
interpretations.

Likewise, the difference between the 'annotation' capability provided by W3C
XML Schema, the language, and XML DTD language per XML 1.0 is the standardized
explicit formal relationship established between items in the schema and their
proper explanations, termed 'annotations.'

Just as with Schema vs. schemas, we will need to distinguish Annotations in
XML
Schema from annotations in general.  'Annotations' in plain English has a
looser meaning. which definitely includes what one does do in XML DTDs to add
explanations and warn about requirements not captured in the formalism.  For
example the DTD comment warning about a requirement for 'name' attributes on
most html:input element types
<<http://www.w3.org/TR/html401/interact/forms.html#edef-INPUT>http://www.w3
.org/TR/html401/interact/forms.html#edef-INPUT>.  The comment to this
effect in
the HTML DTD is an annotation in the common sense.  What Schema provides is a
formal structure for annotations which are thereby stronger language than the
less formal means of binding the additional knowledge to the formal structure
established within the formal language of the schema or DTD.

We will need, in our document, ways to both talk about XML schemas in the
broad
sense, including RELAX and TREX, as well as W3C XML Schema.  I suggest that in
our document we not repeat the W3C throughought but define "XML Schema" to
be a
reference to W3C XML Schema (language)
<<http://www.w3.org/XML/Schema>http://www.w3.org/XML/Schema> and distinguish
"XML schemas" [with two lower-case s'es] to be a reference to the general
class
of schemata constraining or informing XML usage.

Al

>Regards DaveP
>

Received on Tuesday, 24 April 2001 12:44:30 UTC