Mike Taylor wrote:

> How would this be different from ZeeRex or the existing OpenSearch
> description files?  Again, I am struggling to see what problem this is
> trying to solve.

The idea is to describe web based services that are out there -- whether
they be SRU, an HTML form based database such as Copac, a set of static
web pages, or whatever.

I've proven to myself that you can develop a "Description Language" that
will do all the above. Ie, given a 'query', it will allow you to:

   o) create an appropriate url
   o) pull back the response from the server
   o) parse out of the response the stuff you are interested in,
      such as a bibliographic record and/or the number of hits or
      any other arbitrary information.

I wrote a little perl module that reads in an XML description of the
service, pulls back a response from the service and parses out of it
the stuff I need.

There's nothing special about doing all that. However I was very
pleased that my little system allowed me to interact with an SRU
server without having to 'know' SRU. If the query is simple or
always of a fixed form, then neither did it need to know CQL.
Double bonus :-) Of course, the description is very tailored
to my needs and is of probably no use to any one else. That's
fine though. I want to use services how I want to use them, not
how the service designers want me to use them.

So my view of the Description Language is that it is something that
will allow you to "screen scrape" a web based service, no matter what
it is. Others may disagree.

Ashley Sanders               [log in to unmask]
Copac A Mimas service funded by JISC