New PEP: The directive statement

[Tim Peters]

[piet at cs.uu.nl]
> It would be much less of a joke when __future__ would be replaced by
> __features__, I think.

[Neil Schemenauer]
> __features__ is a terrible name.  Its pretty obvious that you
> (and a lot of other people on this list) don't understand what
> __future__ does.

When I first read the Pentium architecture manual, its description of how the
Pentium handled floating-point exceptions was so off the wall I simply
couldn't believe they meant what they said, and went off writing code after
mentally substituting a *reasonable* (but related) meaning of my own.

Heh.  Turns out it meant exactly what it said!  Cost me plenty to recover
from that, too.

I'm not sure we're not suffering a similar confusion wrt __future__:  there
are no exact analogues in any other language I know of, so I wouldn't be
surprised if people think this *is* for "optional features", and that
"optional" is what they'll *read* no matter what the PEP says.  I may like
Martin's "transitional" better for this reason -- provided that it wakes
people up enough to think about what they're reading.  Alas,

    from __transitional__ import nested_scopes

is not only ugly, it's not even funny, so would be worse than "from
__future__ ...".  It works better with "directive" because it keeps it
uniformly bland.

if-you-can't-laugh-at-your-tools-you're-taking-them-too-seriously-ly
    y'rs  - tim