> (There have been some suggestions that ... encoding
> declarations [be] optional if there is an external carrier with a
> character-encoding label...
I hope those sentiments are resisted. Having something like
a declaration that is transport or operating environment
independent seems a lot simpler, reliable, and understandable.
If the declaration is redundant, it is harmless.
> The reason I, for one, didn't lobby for allowing change of encoding
> within an entity...
For what it is worth, I'm in agreement. In practice I don't see
much need for such a feature and there exist straightforward ways
of handling such problems when they do exist.