wrote:
> Another way to look at it is to take the common intersection
> (as you say) of current implementations and standardize that
> set of functions. We can leave it at that. If vendors then
> want to define subsets or supersets of that for special uses,
> then let them. And if something really interesting happens
> with a particular profile/level/package set, then we (or
> someone else) can standardize that in the future.
If an implementation implements only a subset, can they claim to be
conformant to the spec? One advantage of levels is that a very
minimal implementation might claim to be conformant at level 1 and a
more complete implementation could claim to be conformant at level 2,
or level 3, or whatever. Levels or packages would also make it
easier for a spreadsheet to declare what is required to read it (e.g.
"level 2" or "basic plus package blarch").
Tom
