I said:
> > I propose that they do this trivially - if it has a potential
> > namespace name followed by ":", it's the namespace.
>
> And that presumes going through every attribute in the spec and saying
> it might or might not contain QNames? They do need special processing
> after all.
Oh, I'm sorry, I meant only for formula attributes.
They're supposed to begin with a namespace anyway (per spec),
and they need special processing anyway to be used.
So for that special case it makes sense.
Yes, I agree, in the general case doing this
would make things very hard.
> But just to repeat, I do think we need a general position on this
> question -- is it OK to use QNames in attribute content, and if so
> where? -- rather than sweep it under the rug of short-term expediency.
> That was my point in raising the question.
I think that's an excellent question. In general, we shouldn't do this often,
but I think we have to say it's okay for formulas because people
already depend on that. I don't know if there are any other places
where that occurs, though, I think it's quite rare.
It'd make sense to create a list of places where it's okay
(formulas and maybe a few other places), and recommend NOT
using them in the rest.
--- David A. Wheeler
