On 10 Aug 2005 at 11:12, Mike Taylor wrote:
>
> What will happen in practice is that when someone needs to ask for,
> say, A Dublin Core record with some record-level metadata, they will
> just make up a schema that contains these things and a URI that
> identifies it. If several people do this separately, they will surely
> end up with different and incompatible schemes. So it seems better to
> me that we provide some guidance.
>
This is exactly the problem. Given the complexity of the current
proposal (the compoundRecordSchema), I will create my local schema
when I need a specific combination of metadata.
The issue is not that it is too complex for SRU. The issue is that no
server-writer will implement this complexity because it is not a real
local requirement to mix and merge according to the
compoundRecordSchema proposal.
So we will end up with servers offering the conventional record
schemes (dc, marcxml) plus a richer local scheme containing dc plus
some elements from rec and some elements from cld etc.
That is in itself not a problem. Ask for DC and you get DC. But it is
a real real pity that no client can tell from explain that the local
scheme offers potentially a lot functionality and contain a lot of
elements that make it worth for the client to request this local
schema or for the client writer to adapt his client.
A server writer is not likely to adapt the server because to handle
comlex requests that cannot be handled when there is no local
requirement (and he won't see them). A client writer however is more
likely to adapt his client because of the availability of useful
metadata elements.
Theo
|