Well, what are the specific problems with the precedence? Often knowing
the problem better leads to finding a solution.
All I know about precedence is that it has bad karma. It smells of
unreliability. Nothing unusual for responders, that.
But if someone were to write it up as a standards track RFC that more or
less encompassed the present usages, there would be some chance that usage
would move in the direction of conforming to it, and other applications
might take it up. But it is already too widely used to ignore it and hope
it will go away. If you can't beat 'em, join 'em.
sounds more like: if you can't manage to do things well, standardize the
stupidity and claim that you like it that way.
We had a similar experience in Usefor with Mail-Copies-To. We would never
have given it that name if starting from scratch, and we might have made
the semantics a little different. But it was already widely deployed and
more or less consistently used. By including it in our standard, we hoped
to establish what seemed to be the current usage, at the same time
officially deprecating some usages that were already around but frowned on
in the real world.
sometimes you can do this. but in most cases there's a fairly fundamental
conflict between trying to document existing behavior and trying to define
what behavior should be in the standard - people keep getting confused
about what they're doing. so standardizing current practice is often a bad
idea. not always, but often.
Keith