At 09:33 PM 12/21/2003 +0000, you wrote:
>On Sun, 21 Dec 2003, Eliot Christian wrote:
> > At 12:57 AM 12/20/2003 +0100, Theo van Veen wrote:
>
> > >Allthough it's a minor thing I would like the "?" being part of what we
> > >call the base-url.
> > >The reason for this is services can take (private) parameters as part of
> > >the base-url.
>
> > I agree very strongly with Theo's point. I think this is common
> > practice, used for example in OpenGIS Consortium services such
>
>How does this help?  We already have private parameter functionality
>available in as interoperable fashion as possible.

My point pertains to a broader scale of interoperability. Although
there is agreement among current SRU implementors on one way to
handle unknown CGI parameters, the Open GIS community addressed
the unknown parameters problem in another way. I believe this
was a matter of some contention when that spec was negotiated.
(If needed, I can contact the OpenGIS folks and ask someone to
summarize their debate on this issue and how they ended up with
the approach they standardized.)

In the OpenGIS approach, known parameters are either in the spec
or explicitly designated as being in the "URL Prefix": "A URL
prefix is defined as an opaque string including the protocol,
hostname, optional port number, path, a question mark '? ', and,
optionally, one or more server-specific parameters ending in an
ampersand '& '."

We could adopt the OpenGIS approach in SRU and further stipulate
that any parameter key starting with "x-" should not be regarded
as unknown. This would give implementors the ability to include
other parameters in the base URL as well as to include "x-"
parameters where needed. The ability to precisely identify which
SRU parameters are unknown would be fully preserved.


Eliot