Robin Fairbains writes:

> let's not lose sight of what we're discussing, though: we need to know
> if the xml-related efforts can provide us with models of what we want
> to do within latex, or whether it's too restrictive for the full
> generality of what latex does for people today.

My guess is that most of LaTeX admits rational translations to XML
but not fully regular ones, a metaphorical illusion to some pre-20th
century mathematics that is not always well understood.

For example, consider the problem of birationally mapping

             P^2  ---> P^1  X  P^1

(where you know what P^N = N-dimensional projective space is, right?)
I claim that we need to follow something like this procedure with
markup.  I do not have a proof.

But I think that with successive approximations one will find that the
LaTeX we know and love is a complicated categorical limit of XML's.

                                   -- Bill