Re: ACTION to draft the 'copy without namespaces' proposal

On 2014-11-26 14:58, Norman Walsh wrote:
> Paul Grosso <paul@paulgrosso.name> writes:
>
>> Should the local-attributes namespace be anywhere in the XSD? I
>> realize the ##other might take care of it, but should we put it in
>> explicitly?
> I can't think of a way. We don't want to enumerate any particular set
> of attributes, we want to allow any attribute in that namespace, but
> also in other namespaces, so I think ##other covers it.

Right but (pardon my XSD ignorance) would it be an error to add:

<xs:anyAttribute 
namespace="|http://www.w3.org/2001/XInclude/local-attributes|" 
processContents="lax"/>

to the XSD (even though ##other covers that) by way of being
explicit and self-documenting the namespace name in the XSD?


>
>> In section 3.1 after the explicit attributes where we mention
>> "attributes other than those listed above" I think we should have
>> another mention of the local-attribute namespace and another reference
>> to 4.3.
> Yes. Actually, I made "Any attributes in the
> http://www.w3.org/2001/XInclude/local-attributes namespace" into a
> member of that list.


Yep, good.


>> Should we add, say, and xila attribute (for the local-attribute
>> namespace) to the DTD fragment?
> Uhhhh...maybe. I'm not sure how strongly I want to imply that there's
> a particular prefix for them.

No more strongly than we do for the xi prefix.


>
>> The second bullet in 4.3 needs to say that if the xila attribute is
>> assigned the null string, any specification of that attribute on a top
>> level included element is deleted
> Do we? We had that rule for xml:id because an empty string is not a
> valid value for xml:id. That's not the case for attributes generally.
>
> I could be persuaded otherwise, but my initial thought is
>
>    xila:foo=""
>
> puts
>
>    foo=""
>
> on the top-level elements.

Well, that's a good point, but Jirka's email suggested exactly
what I was saying (and recorded in the minutes of our telcon),
so if you don't think we should give people a way to delete an
attribute, it's something we should discuss further.

I see four options:

1.  there is no way to delete such an attribute;
2.  null string deletes the attribute and there is no way to
     set such an attribute to the null string;
3.  null string deletes the attribute and we come up with some way
     to set such an attribute to the null string--like xila:foo="#null#";
4.  null string sets the attribute to the null string and we come up
     with some way to delete such an attribute--like xila:foo="#delete#".

None of them thrill me, but I can't think of another option.

>
> Anyway, I've updated the drafts at
>
>    http://www.w3.org/XML/2012/08/xinclude-11/
>
> and:
>
>    http://www.w3.org/XML/2012/08/xinclude-11/diff.html
>

thanks, Norm.

paul