Why "from __future__" stinks; a counter-offer

[John W. Baxter]

In article <9969ta$2is$1 at newsg3.svr.pol.co.uk>,
 Will Newton <will at nospam.misconception.org.uk> wrote:

> Aahz Maruch wrote:
> 
> > So far, no other person has actually come up with a scheme that addreses
> > the problem at hand, as opposed to simply whining, suggesting minor and
> > worthless variations, or putting forth even *more* complicated solutions.
> > One of the nice things about this solution is that if it is decided to be
> > an ugly wart, it can be easily dropped for Python 2.2.
> 
> How often is this "from __future__" expected to be used?
> 
> Just experimental code written for bleeding edge versions of Python or a 
> significant portion of code (that will eventually become legacy code)?
> 

It seems to be for gentle introduction of changes which are considered 
good (a category in which I would file the nested scopes as it is in 2.1 
and will be), but which will cause code breakage.  If something doesn't 
break code, it can go in without the mechanism.  

Note that no one ever has to write such an import (although folks will 
have to read them)...one can instead elect to wait until the future 
feature becomes *the way things are* before using it.  (But dropping the 
import into existing code long enough to observe breakage in 2.1 would 
seem more prudent than waiting...even without making use of the nested 
scopes.)

And the from __future__ import xxx has at least the advantage that it 
seems to work.

I would expect there will not be scores of those; more like a handful or 
two, since the BDFL has made it clear he would rather not break code.

Meanwhile, the
   from __future__ import xxx
is presently in use for exactly one feature, with no more on the horizon 
for 2.1.

That would seem to leave time for someone(s) who prefer some 
alternative(s) to write (a) PEP(s) proposing same...convincing enough 
and implementable enough, and such input could make
   from __future__ import nested_scopes
the only instance of from __future__.  

But I don't think the proponents of something else should wait very 
long, and I think they should spend their energy on the (er) future, not 
on this instance in this release.  Once from __future__ has been used in 
a couple of releases, changing it would be harder than it would be now 
for 2.2 et seq.  I think Tim's description of the from __future__ PEP as 
darn near retroactive might enhance the chances of an alternative PEP 
displacing it (but not for the nested scopes thing).

  --John

-- 
John W. Baxter   Port Ludlow, WA USA  jwbnews at scandaroon.com