MIVOT: fully qualified attribute names

[Harrison Paul]

> On 23 Feb 2023, at 10:14, Laurent Michel <laurent.michel at astro.unistra.fr> wrote:
> 
> Patricks
> 
> The real problem is that VODML requires 2 elements for naming one thing:
> - No way to make sure they are consistant to each other
> - No rules telling which one has to be used in which context
> (by the way, this gave me the opportunity of 10 days of discussions with Markus :-))
> 
> However the VOMDL spec tells that a name must be unique within its scope.
> - nothing about the way to construct VODML-ID (a good reason not to parse them)
> - nothing about the consistency between names and VOMDL-IDs

The spec might not say anything explicit, but in fact the VODML-ID does have structure in that it is built from the names of an element and its parents. In this sense the name is a redundant piece of information, but it is quite difficult to say in words how to create the VODML-IDs without making reference to the name element at each level in the tree.

> 
> <Screenshot 2023-02-23 at 10.58.33.png>
> 
> 
> The name uniqueness is required but not validated by the schematron, this is a real issue.
> In such a case (and in general), I lean toward using VODML-IDs  which are featured as identifiers 
> 
> My thought is that there is no easy way to  this.
> - If someone feels like implementing the name uniqueness in the schematron: welcome.
> - We could open errata
>     - encouraging the usage of vodml-id
>     - encouraging to check the name uniqueness

I will check, the schematron might already do this - if it does not, I will add a rule to check.