On redefine for minor versioning

[Bryan Rasmussen <brs@itst.dk>]

Hi,
two things I find troubling about the idea of using redefine for minor
versioning:

1. Are we in agreement that a minor version is a version that only has
changes capturable by redefine?

2. I have had problems with redefine in processors in the past, generally
not if there was just one redefine but when there was a good number of them.
So I would hope that using redefine will have a lot more stringent processor
testing than currently. 

Question: If we have minor version 1, redefining certain things from the
major version, if in the next minor version we want to redefine some of
those same things do we redefine from the already redefined or do we
redefine from the original?