Re: [Gzz] PEG: vocabprocess -- cleanup and freeze RDF vocabularies

[Benja Fallenstein]

Tuomas Lukka wrote:
> =============================================================
> PEG vocabprocess--tjl: process for RDF vocabularies
> =============================================================
>
> :Author:   Tuomas J. Lukka
> :Last-Modified: $Date: 2003/05/11 15:57:36 $
> :Revision: $Revision: 1.1 $
> :Status:   Current
>
> It seems that vocabularies are easy to create but difficult
> to define and maintain. We need more process for vocabularies
> which will be put into actual public use.

Agreed.

> This PEG changes the current fenfire vocabulary quite a bit,
> moves a lot of stuff into lava and creates rules on how 
> new vocabulary terms are started and how they end up in the proper
> vocab instead of lava.
>
> Issues
> ======
>
> - Is it ok to have a separate namespace for experimental things?
>
>     RESOLVED: Yes, conversion can be automated / done with inference.

I don't agree. It's still too inconvenient. You should give a reason why 
it is *desirable*, not just that it is *possible*.

>     Any URI in the experimental space should not be widely used
>     before being properly defined and accepted.

This is entirely possible without marking it experimental in the URI.

> Changes
> =======
>
> General
> -------
>
> Create new package, ``org.fenfire.vocab.lava``.
>
> Move most of the vocabulary entries into lava.
>
> Freeze ``org.fenfire.vocab``. Changes only through PEG process

Ok.

> Change the prefix ``http://fenfire.org/vocabulary``
> to ``http://fenfire.org/rdf``.

This looks bad for two reasons:

- Are you positive that we don't want http://fenfire.org/rdf for 
anything else later, such as a webpage about Fenfire's relationship to 
RDF? If this is the root of our vocab definitions, I expect a list of 
our public vocabularies there.
- NEVER use URIs you want to be permanent without a timestamp.

Actually the above could be used with a timestamp, but you don't 
explicitly say so. Please define a full template for namespaces. I suggest::

    http://fenfire.org/terms/YYYY/MM/[DD/]namespacename

where ``YYYY``, ``MM``, ``DD`` are the date (or month) when the 
namespace was created, and ``namespacename`` is chosen by the person 
creating the namespace. (DD is optional.)

Items in the namespace would have the form::

    http://fenfire.org/terms/YYYY/MM/[DD/]namespacename#id

where ``id`` is chosen by the person creating the namespace.

> All new words define without PEG go into org.fenfire.vocab.lava
> and use the prefix ``http://fenfire.org/EXPERIMENTAL/rdf``.

I agree that new words w/o PEG should go into org.fenfire.vocab.lava, 
but I believe they should follow the same pattern as the above. If they 
have webpages at their URIs, these webpages should clearly say the term 
is experimental.

> All entries in vocabulary classes shall have their **official**
> definitions there, in their javadocs. There shall be no members
> or classes without good documentation.

Good.

> Vocabularies
> ------------
>
> ALPH
> """"
>
> Remove ``content``, is in FF.
>
> Remove ``clone`` and ``cloneType`` and ``dataType``, 
> not current/relevant.
...

Um, please do this exactly the other way around: specify what is 
*retained*, giving precise Javadoc definitions.

- Benja