The tail shouldn't wag the dog

The problem of how relative URI references should be treated
is, as anyone not entirely comatose must have noticed, a
vexing one.  It's a little like the toy designed by a
psychologist to teach infants about the nature of the modern
world: any way you put it together it is wrong (or at least
not entirely right).

But to have relative URI references -- an uncommon though
not nonexistent case -- drive the entire namespace mechanism
seems to me to be a great mistake, a case of the tail
wagging the dog.  I think it's important to get the common
case right first -- and although views may vary as to what
"right" is, I think that's the literal interpretation as it
is in the current version of the namespace spec.

Once we've fixed our interpretation of the common case, we
can then set it aside while dealing with the uncommon case.
Personally, I like James Clark's suggestion of using a
standardized base for relative URIs.

Paul Abrahams