At 06:02 PM 12/24/2003 +0100, Theo van Veen wrote:
>That's what the majority wants, but we need a way to solve the problem
>of the "base-parameters".

... i.e., to define what are regarded as "unknown" parameters and
what is the proper behavior of a server when encountering an unknown
parameter.

I still like the idea that an SRU server should simply ignore any
unknown parameter. HTML has that behavior and I think it has proven
to be a source of much interesting innovation. The downside, of
course, is less precision in giving feedback to fat-fingered typists.

Quite separate from server behavior is the need to define "unknown".
I guess an "unknown" SRU parameter would be any parameter having a
key that does not start with "x-" and also does not match any of the
"searchRetrieveRequest parameters". (I don't know where to find a
complete list, but the known parameters include "query", "sortKeys",
"startRecord", "maximumRecords", "recordSchema", ...).

It may be wise to note that that the set of searchRetrieveRequest
parameters is unordered. It should also be noted that none of the
known parameters may occur more than once in a query string.

The case of "no parameters" (i.e., base-url + "?") is defined to be
the SRU syntax of an explainRequest. It would be nice to require
SRU servers to accept a base-url without an "?" as a valid equivalent.

I also think it would be nice if SRU servers were to accept a query
string that does not terminate with an "&" as a valid query string
(i.e., the end of a query string implicitly terminates the last
parameter value).