LeVan,Ralph wrote:

>Braindead clients forget things, like the explain response that they just
>had in their hands.  The response to a search does not contain the name of
>the database that was searched.  It does not contain the list of indexes
>that might be used to refine the search.  It does not contain the list of
>alternate schemas that might be used to retrieve any records.
>
>
Sorry to blend into an ongoing discussion, buuut I do not think that a
standard should
aim at beeing good for brain-dead clients. I feel, it was better beeing
good for normal clients, and good for saving bandwidth, ressources, and
programming effort in the server side.

>But, if the base URL had been returned, the braindead client could have
>asked for the Explain record again and incorporated that data into the
>search response.  (There is a method in XSLT for going off and getting
>another record dynamically and using it as part of the original record.)
>
><Warning>Cache your explain records!  They are going to be asked for after
>every other type of interaction.</Warning>
>
>
>
This is exactly what some servers can not do. I am just writing a PHP
SRW frontend to a  SQL Oracle backend, and caching is the last thing I
want to do, given the possibility to get a new tread/child of apache
every time my server is hit by a HTTP/SOAP/SRW
request.

This said, there might be quite many other good uses for the base URL,
but the existence of braindead clients is not such a good reason.

IMHO ...

cheers, Marc Cromme, Indexdata