Status: RO
[Continuing to cross-post to dc-international, because
Andrew did, and I know Tom Baker is not on the dc-datamodel
list. Apologies to those with strong objections to
cross-posting]
Andrew,
You wrote:
>...
>My preferred approach is your Approach 1 (i.e. separate files for each
>locale with a locale switch file).
Good. I was leaning that way myself.
> Separate files were implemented
>both in Reggie and PrismEd (a side effect of having configurable
>schemas!), but I never got around to implementing the locale switch
>file. Note that for Approach 1 to work, the URL in a namespace
>definition must point to the switch file, not to a locale. For
>robustness, each locale should also point to a locale switch file so
>that if an editor is pointed to a locale file (by mistake) it can
>recover.
Yes. I wonder if it would be best to make the namespace definition
point to a file containing the language-independent parts of the
schema *and* a pointer to a separate switch file.
When I say "file" I mean of course "URI". And "file containing..."
means "URI which when dereferenced would return...". It's just so
much easier to think in terms of files.
>...
>You claim of Approach 2 (i.e. mixing the locales together in one file)
>> The data (i.e. schema) is easier to manage, because you don't have to
>> worry about putting it in separate files. It could be stored in a
>> database.
>
>I disagree; I believe that Approach 2 makes it *harder* to manage the
>various locale information because only *one* organisation is able to
>manage the schema information. That one organisation has to manage
>everything: from the default schema to the Ethopian Dublin Core schema.
>Approach 1 allows the management to be decentralised.
OK, I'll back down on this one.
What I meant was that one of the attractions of Approach 2 is that
it is easier to define and discuss it by drawing RDF nodes and arcs,
and you don't have to worry about "files". But eventually you have
to deal with the files anyway, and then Approach 1 is better.
>For completeness, you also need to specify what schema properties can be
>expressed in different languages. In Reggie and PrismEd there are two
>"HCI" properties:
>* Label. This is a short (often one word) description intended
> to label buttons etc.
>* Description. This is a longer textual description intended for
> 'help' screens. This text was marked up in HTML and was pointed
> to by a URL from the schema.
When Tom Baker brought this up, I told him he could just define
such property types in his own namespace. The report from his
meeting last month indicates that the DC-in-Multiple-Languages
working group would prefer them blessed by the DC community.
*You* could define them in a namespace that you control. That
will work just as well as if they were in a namespace controlled
by the W3C or OCLC. In fact I think you and Renato are
particularly well placed to define property types for use with
general purpose metadata editors (I don't know if you are
working together, but you mentioned Reggie). I don't think it
is really a Dublin Core matter, because the DC is just a
vocabulary for metadata about real resources.
You could also define the property types for the switch file.
The DC-in-Multiple-Languages group may make use of them (the
report mentioned experimental implementations) and tell you
of any difficulties.
Regards,
Charles [log in to unmask]
------------------------------------------------------------------------
Any views expressed in this message are those of the individual sender,
except where the sender specifically states them to be the views of
Reuters Ltd.
|