[tei-council] Conformance draft: namespace purity

Tue Apr 3 22:05:41 EDT 2007

Good morning, Christian!

> > I think that James' stipulation in the draft that new elements
> > *must* be in a a non-TEI namespace might be a bit overstated --
> > perhaps we should *permit* null-namespace additions.
> 
> This seems like splitting hairs, but since the null-namespace is
> certainly not a TEI namespace, I would think this is licensed by
> the current draft.

I had thought that an element w/o a namespace declaration in force is
not really in a namespace at all, rather than being in the null
namespace (which we were using as shorthand notation for this
condition).

> This is not what you will do. You will add a
> xmlns="http://www.tei-c.org/ns/1.0" to the TEI element above and
> then get the following:
>     <text>
>       <body>
>         <head>Blue</head>
>         <p>My encoded text with <thingy xmlns="">the thing</thingy> I am
>         interested in encoding</p>
>       </body>
>     </text>

I had forgotten about the possibility of xmlns="" -- it is the same
as no namespace, right?

> The downside of this of course is that it will need to jump some
> through some hoops in DTD land where you have to emulate namespaces
> with prefixes and or #FIXED attributes.

I'm not sure if I want to use that as an argument against requiring
added elements be in a different namespace, or against supporting
DTDs! :-)

> The annoying part starts only if you want to process this, 

I think it's pretty annoying to edit & proofread, too.

> ... but even there you could argue it is entirely reasonable to
> have to deal with your own stuff yourself and this namespace
> business helps you separate your stuff from their stuff.

Absolutely. But my major point is that you are going to have to deal
with not only your own stuff, but TEI stuff anyway. It's not like
it's really possible to have out-of-the-box TEI software to process
your TEI documents. Besides, if such a beast existed, it would likely
choke as soon as I use a customization that is not a restriction, but
since it isn't a new element, doesn't change the namespace. (E.g., if
I were to change a content model to permit an element that isn't
available in vanilla.)

G'night!