[tei-council] non TEI attribute namespace

[Martin Holmes]

I think it would be better for interoperability and clarity if we 
required (or strongly advised) that all custom attributes be explicitly 
declared in a different namespace. For processors, this means that all 
unprefixed attributes, along with those prefixed attributes that are 
part of the standard schema, are known to be TEI, and all custom 
attributes can easily be distinguished because they are neither in the 
xml namespace nor in the empty namespace.

It also makes it easier to group custom elements and custom attributes, 
because they'll presumably be in the same namespace(s).

Cheers,
Martin

On 14-03-04 07:05 AM, Lou Burnard wrote:
> Alexei Lavrentev, whom some Council members know, asked me a good
> question when I was teaching ODD in Lyon last week.
>
> "What I did not understand is whether adding a zero-namespace attribute
> to a TEI element
> (e.g. lg/@asonancia) is
> a) ok, if the attribute is declared in the ODD ;
> b) not forbidden but is not a good practice ;
> c) is not conformant. "
>
> To expand on this a bit, all TEI attributes are in the empty namespace
> and so there's no formal reason why my ODD should not declare a new
> attribute and place that also in the empty namespace. Do we think that
> that is perfectly OK?  It will of course be valid according to the
> schema my ODD generates, but may upset other TEI processors which are
> foolish enough to expect to see only TEI-defined attributes.
>
> Of course, I can be squeaky-clean  by placing my new attribute in my own
> namespace, and I will be required to if it has a name already used by
> the TEI. But if neither of these is the case, am I still TEI conformant?
>
>
>
>