A counter-proposal to __future__ in PEP 236

[Tim Peters]

[Sverker Nilsson]
> I think the directive directive is better too. When I read about
> the future import I was worried over what the new semantics
> of import would be, for example what would happen to the
> identifier that would be imported: would it be visible,
> probably not but that's a new semantics to understand & learn.

No, I think PEP 236 is quite clear that, at runtime, future_statements *are*
utterly vanilla import statements.  They do import module __future__.py, and
they do import the feature names (which are bound to detailed release info,
which is important metadata for people who use this; the "directive" proposal
loses that).