Well, Mike and I agree about as often as he and Theo do, so this is
going to be a red letter day!
First, I don't think we should go down the path of creating a schema in
which we encourage folks to dump whatever useful stuff they can think
of. It smacks of the bib-1 attribute set.
Second, I don't think that the semantics of new elements dropped
randomly into a schema are any better understood than random elements
taken from any other schema. In other words, I don't think that the
newly added dmd:ralphsStuff element will be any better understood in a
DC record than just dropping onix:price would be. Just because the
element is defined in our ever-changing schema does not mean that it
will be understood by the client. And please, don't suggest that schema
versioning will solve that problem. It is just namespace mangling and
does nothing to help interoperability.
Now, here's where Mike and I probably stop agreeing. I think we should
just go ahead and do what Theo asks. He's already said that ONIX is not
an acceptable choice for him because his usual clients don't know what
to do with it. That said, they probably won't know what to do with an
onix:price element in a DC record either. But they probably won't blow
up and Theo has reason to believe that the clients will just display the
data to the user who WILL recognize it as price information, even if the
client doesn't.
There is NO easy solution for incorporating unexpected data into an
existing schema, other than hoping that the clients are smart, robust
and flexible. The growing use of microformats in all sorts of records
depends on that and I think we should too.
Ralph
> -----Original Message-----
> From: SRU (Search and Retrieve Via URL) Implementors
> [mailto:[log in to unmask]] On Behalf Of Ray Denenberg
> Sent: Thursday, April 03, 2008 12:46 PM
> To: [log in to unmask]
> Subject: Re: SRU Record Metadata and DCX Schema (was: Price
information
> in SRU responses)
>
>
>
> From: Mike Taylor
> >Finally, there is the proposal that you posted, namely the ability to
> extend Dublin Core records with elements from the Record Metadata
> namespace.
> I think this is a good and useful proposal, but it has nothing to do
> with
> the particular problem we set out to solve.
>
>
> I think you're still not grasping what I'm proposing.
>
> I'm NOT proposing to extend our DC schema with the RMD namespace. I'm
> proposing to extend the DC schema with both the RMD namespace AND the
> DMD
> namespace.
>
> The DMD namespace will be a discrete set of descriptive metadata
> elements
> that we come up with collectively (just as we did for the DMD
> namespace) one
> of which will be 'price'. Note that extending DC for RMD was a one
> line
> addition to the schema <xs:any namespace="dmd> , and extending it
> further
> for AMD will be another one line addition <xs:any namespace="DMD">.
>
> This approach does not have the interoperability problem you cite
> because,
> dmd:price for example, will be a well-known element from a well-known
> namespace.
>
> If the person who started this discussion, about price, wants to use
> ONIX,
> fine, but at least we'll have this in place for the next time someone
> has a
> similar need.
>
> We can simply call this schema DCPlus.
>
> And I can trivially create the schema that Theo asks for (DCX)also
> with a
> one-line addition <xs:any namespace="any"> (or something to that
> effect).
> Which I'll do as soon as I'm back in the office (I'm away for a few
> days).
>
> --Ray
|