[Ex-plain] Latest DTD (was ZIG)

Mike Taylor mike at tecc.co.uk
Fri Mar 8 17:52:05 CET 2002


> Date: Fri, 8 Mar 2002 01:13:03 +0000 (GMT)
> From: Robert Sanderson <azaroth at liverpool.ac.uk>
> 
> Executive Summary of my opinions:
>   * indexType and index concepts need documentation
>   * elementSet needs a title
>   * Do elementSet and index need description?
>   * <use> etc are descriptive for ease of record creation
>   * sortKeyword represents 'private sort key'
>   * Any text element can be repeated with different languages
>     (eg title, description, history, extent...)
>     As such the 'if no preference, use this' should have primary="true"
>   * recordSyntax/name and attributeset should be OIDs, or a rigid list of 
>     equivalent keywords. (eg BIB1, GRS1, XML)
>   * We need a date format. I suggest unix ctime.
>     Eg:  "Thu Mar  7 20:15:17 2002 EST"

Beyond those already discussed, here's a summary of _my_ opinions:

    * It's time you sent me v1.2 of the DTD!

That's it :-)

More or less ...

I vote "no" on the "description" for elementSet and title, on the Less
Is More principle.  The same basis on which I oppose Rob's <indexType>
and Alan's whatever-they-were-called thingies.  Remember folks, KISS!

> This is useful for ZING as well, if I recall correctly, as they
> /only/ understand indexes.

... that's SRW, of course ...

(I feel like RMS now, constantly bleating on about "GNU/Linux" :-)

> <indexTitle lang="australian">Awwwwtha</indexTitle> ;-)
> 
> (Couldn't resist, I'm from New Zealand originally =) )

:-) :-) :-)

> > Sorry, I don't understand exactly (and I might not have been
> > clear).  By enumerating attribute types (which are numbers in
> > Z39.50) as elements in the DTD, it means if anyone ever comes up
> > with a new attribute type, the DTD needs to be extended. This
> > seemed architectually bad to me.
> 
> Yes. If someone were to come up with a new attribute type, then we
> would need to put in a new tag to cope with this.

I agree with Alan that this is unpleasant and half-baked.  I would
prefer to see numeric attribute types for several reasons: they map
more closely to what we are describing; their meaning does not not to
be interprered differently in the context of different attribute sets;
and of course extensibility.  Not least when people/projects invent
their own new attribute sets with new attribute types, which is a
perfectly legitimate thing to do.

> However considering how many implementations of the attribute
> architecture there are, I think we'd have enough fore warning to add
> and implement this without breaking a sweat :)

Oi!  What about all them Zthes servers?!

> I think people are more likely to be able to hand craft the named
> attributes than the technically superior numeric definitions, and
> this is something that we need in order to get widespread
> acceptance.

I don't really see this.  If, as we're anticipating, 99% of the
records created are describing BIB-1 searches, then they'll be created
by people whose typing fingers' device-driver neurones know full well
that 1=use, 2=relation, etc.  I don't see this as a remotely
significant barrier to implementation.  ("Well, I _was_ going to put
an Explain-- serber up, but then I couldn't remember the BIB-1
attribute type for trunctation, so I thought, naaaah.")

> > > > * What are legal format names? (What is the exact format etc)
> > >
> > > The OID, or the official name for it?  GRS-1, SUTRS, MARC, XML,
> > > SGML, etc
> > 
> > I guess the OID would make the most sense (silly me). It saves
> > having to define standard tables - and is extensible without
> > changing the Explain-- spec.
> 
> OIDs are correct, but I'd really like to see either OID /or/ some
> very precisely defined list of equivalents.  'XML' is much easier to
> write than '1.bla.bla.bla.bla.bla.bla.bla.105.9'

I agree with Rob.  We should support all OIDs (for extensibility), but
also specify a small number of widely-used symbolic names for the
record syntaxes that crop up all the time.  Or, maybe better, use the
MIME-types -- "text/xml" and suchlike.

 _/|_	 _______________________________________________________________
/o ) \/  Mike Taylor   <mike at miketaylor.org.uk>   www.miketaylor.org.uk
)_v__/\  "``I want to see the zoo,'' she said, ``I want to see the
	 zoo.''" -- song from Mike Leigh's "Nuts in May".





More information about the Ex-plain mailing list