Here is a new overview with Mike's, Theo's and Bill's second proposal
added.
Some proposals require a new version, some don't. With so many
solutions we carefully need to describe the actual underlying problem. I
will be on holidays next week; have fun :-)
---------------------------------------------------------------------------------------
Rob:
---
Modify recordSchema semantics to allow multiple schema
identifiers/names.
If multiple schemas are given:
a) The first MUST be a wrapper schema, the rest are schemas to
include.
If the first isn't then it's an error. (Ray)
b) If the first is not a wrapper schema, then it's a choice of schema
(as per one of the current extensions) (Rob)
Mike:
----
1) Just define new identifiers for each combination of wrapper and
wrapped
schema.
2) forget it
Theo:
----
1) Allow multiple record schemas in recordSchema but only allow a
single
wrapper schema, to be specified by the protocol.
2) Support one specific schema/application profile (DCX) for the
specific case of
"qualified dc + rec + collection descriptions + service descriptions"
Matthew:
-------
Allow compound schemas to be identified in a new nested XML structure
in
SRW, and map that structure into a set of parameters in SRU.
Ralph:
--------
No new identifiers for local mixes. Local identifier for local
wrapper
and some way to hint at contents in Explain.
Bill:
-----
1) If you require a record in multiple recordSchemas a separate
transaction to obtain the record in each recordSchema.
2) adopt the approach taken by OAI PMH and place record information in
a record "header"
separate from the metadata.
|