I was one of the people who asked for textual identifiers for schema
properties and classes. From a developer's point of view, purely
numeric identifiers for classes and properties is hell... it means
constant lookups against documentation, queries and data that look
like soup, really slows you down and prohibits communication and
learning. By analogy, imagine working with a database schema where
every table name and column name is just a number. Schema property and
class identifiers need to be memorable.
On the other hand, I think it's fine for SKOS-like and other "value"
vocabularies to use numeric identifiers. "Value" vocabularies are used
quite differently from schemas. Generally, within code, there is much
less direct use of specific values, and therefore the developer has
less need to memorise them. Within user interfaces, users interact
with these vocabularies via their textual labels, and so are
indifferent to URI form.
So I'm happy with the way the registry does URIs right now. Of course
the registry could introduce an option allowing vocabulary owners to
choose the form of their URIs, but I would still encourage schema
owners to use textual mnemonic ids, and value vocabulary owners to use
numeric ids, as best practice.
Cheers,
Al.
On Mon, Nov 17, 2008 at 02:09:31PM -0500, Diane I. Hillmann wrote:
> Karen:
>
> It's a quick fix. Once you change the name, the system will prompt you
> to change the URI, and when you say "yes" it will do that and keep track
> that you've done so.
>
> Yes, well, we changed the schema side to textual from numeric because we
> were criticized for that early on. Can't please everybody, I guess ...
> :-)
>
> Same with roles--we had some discussion about it (which I thought you'd
> been involved in but perhaps not) and felt it would be clearer if they
> were separated. I still think so, actually, but there you are--another
> six of one, half dozen of another kind of decision.
>
> Diane
>
> Karen Coyle wrote:
>> On Mon, Nov 17, 2008 at 8:30 AM, Alistair Miles
>> <[log in to unmask]> wrote:
>>
>>> Yes, but I expect all property names to start with a lower case
>>> character -- this is a minor issue in the RDA elements vocab, the
>>> property URI just needs changing from .../Publisher to .../publisher.
>>>
>>>
>>
>> I'll fix that, but I need to know (from Diane or Jon) if I can change
>> the record "in place" or need to delete and re-create. The question is
>> whether the versioning will work if I change the identifier.
>>
>> BTW, this is an illustration of why it might be best to use
>> non-textual identifiers. The decision was made to include the name of
>> the element in the URI, but it makes me uneasy. (I also would like the
>> RDA roles and elements to be in the same element list... the way it is
>> I think we proliferate the confusion about the roles and what they
>> mean. But for the moment we probably need to keep track of them
>> separately since RDA does.)
>>
>> kc
>>
--
Alistair Miles
Senior Computing Officer
Image Bioinformatics Research Group
Department of Zoology
The Tinbergen Building
University of Oxford
South Parks Road
Oxford
OX1 3PS
United Kingdom
Web: http://purl.org/net/aliman
Email: [log in to unmask]
Tel: +44 (0)1865 281993
|