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
recordSchema with a fixed name ("caption", "brief" or whatever) and I
agree that this can be used in a generic way. I would say title should
be mandatory (even when it should be constructed as "no title") and
recordid, author and date mandatory if applicable.

As we are discussing the transformation of the ouput anyway I propose
we also provide the possibility of using references to an xslt-xPath
file. Currently at KB we have implemented this as a site specific
parameter and use it as part of the SRUbaseURL.
We now support the parameters:
xsl= to specify the stylesheet reference to be put by the server in the
XML header
xslt= to specify a local xslt file at the server (these filenames are
ofcourse unknown to external clients)
Because xPath is not really useful as URL parameter I expect we will
provide a reference to an external xslt file, but I do not yet know the
security issues involved. It would be nice to incorporate this in the
SRU/SRW standard at the same time as xPath.

Theo

>>> [log in to unmask] 8/7/03 1:15:20 nm >>>
> In general we provide a standard in which a client requests A and
gets
> A or a diagnostic. I prefer the server to provide A or a diagnostic
plus
> B, where B is not just "doing their best to provide a useful
answer"

But as Matthew has pointed out, you can't get a diagnostic plus B as
there
isn't space for a 'surrogate' diagnostic and a record.

> In the xPath discussion A means "records conform the specified
xPath"
> and B means "short DC records". With or without xPath I would like
the
> availability of short records (like brief in Z39.50) to be returned
on
> request or in combination with an diagnostic when a requested schema
is
> not available. I would like to postpone the discussion on "what is a
> short DC record?".

So what you really want (what you really really want) is more
[meta]data
in the surrogate diagnostic, so even though it is a diagnostic, it's
got
some information about the record it's standing in for?

I'm not completely adverse to such a proposal, but the fields would
need
to be nailed down as to exactly what can be returned. I would think
unique record id, title, author, date would be sufficient for most
purposes.

This wouldn't apply to just xPath surrogates, but could be used for a
lot
of them. If you know there was a record there and it has subsequently
been deleted, chances are good that you can at least give a docid.

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