MIVOT: fully qualified attribute names

[Markus Demleitner]

Dear Laurent,

On Thu, Feb 23, 2023 at 03:24:45PM +0100, Laurent Michel wrote:
> > I think that if we say that the VODML-IDs have structure, then we
> > can allow Markus’s desire for non-qualified attribute names as it
> > is possible to work them out with simple string manipulations.

No, that has not been the reason of my qualms with vodml ids.  It's
been that they needlessly complicate everything without giving a
tangible benefit (just spent another ~4 hours of implementation time
on them on Monday).  Plain attribute names work just dandy everywhere
outside of VODML and MIVOT, and they would work perfectly well for us
as well.  String manipulations absolutely don't enter into it.
Having to manage and parse VO-DML does.

No, I gave up on that discussion because contrary to my previous
impression, it is at least *possible* to work with vodml-ids.  It's
just annoying, and annoying isn't enough to block the whole thing.

> This discussion strengthens my conviction to require MIVOT working
> with the VODML-IDs. They are identifiers, better controlled and
> managed,  thus safer to use.  Anyway, this MIVOT topic is closed
> after Markus has proposed a little spec update.

Well... you see, there are many reasons why deprecating vodml-ids
entirely would be a good thing; one is that DM identifier ("prefix"),
class name, and attribute name are perfectly enough.  As Pat says:
It's hard enough to come up with *one* name, let alone two.  And as
Pat has noticed, when you have two things doing the same thing, at
least one of them will eventually be wrong.

While I won't push such a deprecation along, I'd certainly support
anyone who would.

        -- Markus