On Thu, 7 Aug 2003, Theo van Veen wrote:
> The diagnostic does not have to be a surrogate for a specific record as
> the request was also not for a specific record. So it can be at record
> level. With respect to the "minimal" metadata I would prefer a
I'm sorry, I don't understand. Surely a diagnostic at record level is a
surrogate diagnostic?
> recordSchema with a fixed name ("caption", "brief" or whatever) and I
You can't have a record and a diagnostic in the same space. We need to
extend the diagnostic schema to allow for some bits of optional metadata
(assuming that we do anything). However you're perfectly able to create a
record schema and ask for it right now.
I'm against having any mandatory fields though otherwise you get issues
like how to specify the title of a record that has been permanently
deleted.
> As we are discussing the transformation of the ouput anyway I propose
> we also provide the possibility of using references to an xslt-xPath
That's one of the proposals already for 1.1 and for SRU or SRW I think
it's useful, though obviously more so for SRU. Either protocol could
benefit from having a link to a style sheet which governs how to display
them. Though (again, I think, obviously) there's no guarantee the client
will use it.
Mike:
> And, by the way, the lesson of Z39.50 (the BIB-1 diagnostic set
> vs. the Diag-1 diagnostic format) is surely that implementors want
> _simpler_ diagnostics even when they are plainly inadequate. So I am
> actually not even convinced that rich diagnostics in SRW are a good
> idea. We already have one capability discovery mechanism (ZeeRex),
> why add another redundant one?
I agree that we shouldn't duplicate information which is available in
ZeeRex, but it seems acceptable to allow more detail in surrogate
diagnostics about the record it's standing in place of? This isn't
something that you can put in Zeerex, or in fact any place other than a
surrogate diagnostic?
For example, without any silly analogies:
Request is for a search on author, xPathed to retrive the first paragraph.
One record doesn't have any paragraphs, it's just metadata. Currently
you'd get back a diagnostic saying 'xpath is invalid for this record'. But
the record is still there, the server knows about it, but the client has
no way to know how to get to it.
Even if we don't include any data from the record, we should at least
(IMO) allow an identifier for the record so the client can uniquely
retrieve it and do its own processing if it wants to.
In fact this is probably enough, then the client can retrieve the record
itself using whatever record schema it finds most useful rather than the
server trying to guess what fields the client will want.
In Z, this isn't necessary as result sets are mandatory. In SRW/SRU
they're not.
Rob
--
,'/:. Dr Robert Sanderson ([log in to unmask])
,'-/::::. http://www.o-r-g.org/~azaroth/
,'--/::(@)::. Special Collections and Archives, extension 3142
,'---/::::::::::. Nebmedes: telnet: nebmedes.o-r-g.org 7777
____/:::::::::::::. WWW: http://nebmedes.o-r-g.org:8000/
I L L U M I N A T I
|