On Tue, Feb 05, 2002 at 02:30:57PM -0500, Ray Denenberg wrote:
> >From the recent discussions I've added three new issues (7-9) at:
> Please tell me if there are issues missing from the list.
There was the recent proposal from someone, mutilation of original
proposal by someone else (me!), and the rejection from yet someone
else. So I am not necessarily proposing we do it, but might be worth
recording the suggestion which was to...
Include an additional lump of XML in the request and response packet
which was implementation specific. Servers can ignore it, clients
should not mandate that requested information be returned. But it would
allow an extensible way of addining information to requests (for me
I could insert a term highlighting request) and responses (could return
a query break down with term frequency information). There would be
a single standard root element with multiple child elements in different
namespaces holding all the different local extensions. Namespaces would
be used to avoid collisions in interpretations of element names.
The rejection (as I recall it) was based on wanting to keep SRW simple
and avoid extensions that could cause interoperability problems later.