[tei-council] attribute names (was MD chapter revised: namespace rules)

[Lou Burnard]

Wittern Christian wrote:
> Daniel O'Donnell wrote:
>> On Sun, 2007-04-15 at 10:32 +0100, James Cummings wrote:
>>  
>>> Sebastian Rahtz wrote:
>>>    
>>>> James Cummings wrote:
>>>>      
>>>>> Yup, I understand and agree with all of that.  So by suggesting 
>>>>> that people use namespaces for their new attributes we encourage 
>>>>> the good practice that TEI already follows of not having 
>>>>> @myspace:where mean something on one element and @myspace:where 
>>>>> mean something else on another element.
>>>>>         
>>>> what if add the TEI-consonant @type to an element which doesnt have 
>>>> it already?
>>>>       
>>> I believe consensus was that this was an 'unclean' change, and thus 
>>> element and attribute are put in a user-defined namespace.
>>>     
>>
>> Indeed it was the consensus. And indeed if you added it to new elements
>> because you wanted to have a global type element, we seemed to think
>> that all types were my:type.
>>
>>   
> Well, things are moving so fast it becomes difficult to keep track.  
> Say I want to add @type to p.  Does this mean my tei:p has to be 
> changed to my:p?   This seems quite drastic to me.  It would be 
> sufficient in my book simply to say @my:type  and leave p as tei:p.
>
> Christian
>
>
Well, as regards namespace, the options seems to be:

1. The line of least resistance: say that your schema is not in any 
namespace (by saying ns="" on your <schemaSpec>)
2. The odour of sanctity: say that the attribute is new (by saying 
ns="http://myspace.com" on the <attDef> in the <elementSpec> in your 
<schemaSpec>)
3. Political correctness gone mad: say that the element is new (by 
saying ns="http://myspace.com" on the <elementSpec> in your <schemaSpec>)

Of these three, I believe (but I'm not sure) that (2) and (3) are both 
what we are currently calling "conformant", but obvious (2) makes more 
sense.

We don't have a name for (1) yet.