> Date: Tue, 16 Dec 2003 12:10:55 +0000
> From: Robert Sanderson <[log in to unmask]>
> We thus also need a 'profiles' section in the documentation.
Yes. Just as the Z39.50 Maintenance Agency has separate lists of
attribute sets and profiles.
> > http://zthes.z3950.org/srw/current.html
> > All comments are welcome.
> Something you weren't to know, but I'm going to migrate all my SRW
> stuff into srw.cheshire3.org for 1.1 (as o-r-g
> (online-roleplaying-games!) is hardly the most appropriate place for
> it (with the possible exception of the L5R database, I guess))
Well, srw.o-r-g.org seems fine to me -- everyone reads the little-end
first, right? Still, if you want to move the documentation, that's
your prerogative. However --
> So can you relink the record metadata set as
> (even though that link says that the identifier is still the old one,
> I'll update that today as well)
You ought to think long and hard before changing the URI of an
established object such as the Record Metadata context set. This is
exactly analogous to changing the OID of the Utility Attribute Set --
not something that would be done lightly. Remember that these URIs
are opaque identifiers, and that people really shouldn't be peeking
inside to see their structure.
I wonder whether I ought to ram that point home by making the URI of
the next context set I create something like
We may not be helping people in the long run by choosing URIs that
_appear_ to be related to the object whose identifier they are.
> WRT the Explain section, I have two responses.
> Answer 1:
> ZeeRex does support saying that the database supports a profile in
> 2.0 with <supports type="profile">zthes</supports>
Cool. I'll update the Zthes profile, then/
> Answer 2:
> If you support the ZThes schema, and all the indexes, then you
> support the profile. All of this can be expressed in ZeeRex
But no application wants to have to check all that stuff separately in
order to check whether a database is a thesaurus. <supports> is a
good answer, though.
(Also, come to think of it, the current Zthes-for-SRW profile stupidly
Rather than indicating that the ZeeRex specifications
should be expanded to include Zthes-specific
information, this deficiency should be construed as
evidence that ZeeRex ought to include some mechanism
for including profiled extension information.
Which is pretty dumb when you consider that _I_ wrote the section
on application-specific extensions to the ZeeRex record. D'oh!)
/o ) \/ Mike Taylor <[log in to unmask]> http://www.miketaylor.org.uk
)_v__/\ A song, a smile, and a recursive-descent regular-expression
Listen to my wife's new CD of kids' music, _Child's Play_, at