[Nml-wg] Identifiers

[Jason Zurawski]

Freek/All;

On 11/8/10 10:45 AM, Freek Dijkstra wrote:
>
>>> Question 2. What attributes to use for references in XML?
>>> a) existing id and idref in NM-WG namespace
>>> b) redefine id and idref in NML namespace
>>> c) create dedicated namespace for just id and idref
>
> Jeroen van der Ham:
>
>> I thought we decided on B?
>
> Jason Zurawski:
>
>> I think b) makes the most sense; we do this in NM/NMC now.
>
> Freek Dijkstra:
>
>> No big prefence. In order: a, c, or b
>
> Talking in private to Jeroen (today) and Jason (at OGF), my question
> seems ambigious. Let me try again.
>
> Let's assume we will make a derived Ethernet schema,
> http://schemas.ogf.org/nml/ethernet/2015/05/
>
> Question 2b. What attribute to use for references in XML?
> a) id and idref in NM-WG namespace
> b) id and idref in NML base namespace
> c) id and idref in NML Ethernet namespace
> d) id and idref in OGF namespace (created for just these attribs)
> e) about and resource in RDF namespace
>
> Am I correct that our preferences is:
> Jeroen: b
> Jason: c
> Freek: a, d or e
>
>
>> To me it seems very weird to reuse NM-WG namespace for something
>> trivial as this and for that reason almost ridiculous to create a
>> dedicated namespace for that.
>
> At the risk of making flamebait:
>
> Redefining the same thing over and over and over again (option c), seem
> to nullify the use of namespaces.
>
> For one thing, the power of XML or RDF comes from using libraries that
> understand some part of the XML, for example about ordering, namespaces,
> and identifiers. (To me that was the compelling reason to like RDF; the
> semantic stuff is only a side-effect for me). By defining this once, it
> is easy to augment a library to treat this as special.  If we decided to
> redefine this over and over again, each time in a new namespace, the
> implementor must write the same code again and again (especially if a
> good XML library is used, where equal name in different namespaces are
> considered something different). I dislike that.
>
>
>> It is uncommon to see [namespace prefixes for attributes]:
>>
>>     <ns:element ns2:attribute="something" />
>>
>> But it is possible.
>
> Yes, it is indeed uncommon for XML. It is common for RDF/XML.
>


If it matters that much for your use cases and you can't rationalize not 
re-using attribute definitions, do what you think make sense.  As 
denoted in my first response, attributes are almost always associated 
with the element because there is never a need to see them re-used or 
sourced from alternate namesapces.

Remember that we are just talking about a simple name and value pair 
(string datatype, nothing complex).  The semantic meaning behind id and 
idRef would remain the same no matter what namespace you are referring 
to.  I doubt there would ever be a case where an NML-base idRef pointing 
to an Ethernet id would be treated any differently than other id/idRef 
situations.  If you forsee the relationship between ids and refs of 
different namespaces being meaningful, please give some examples.

The added complexity to simply reuse the following text:

> Identifier =
>         attribute id { xsd:string }
>
> IdReference =
>         attribute idRef { xsd:string }

Hardly seems worth the trouble.

-jason