Booleans, integer division, backwards compatibility; where is Python going?

[Guido van Rossum]

Donn Cave wrote:
> I sure fail to see the point of this middle ground.  On one hand, you
> have a language that is manifestly very useful to programmers in a variety
> of disciplines but fails to be the perfect language.  On the other, you
> have a vision of the perfect language, which in principle ought to be
> more useful but definitely will break some significant amount of code
> written today.  In between, you have all the problems of both - breaks
> code today, doesn't prevent tomorrow's breakage, and isn't the perfect
> language anyway.  Major Python heads fight tooth and nail against changes
> that they might well accept without complaint as features of a really
> overhauled language.  Developers turn away, platforms like Redhat who
> could ship Python are put in a difficult position.  In the present release
> schedule, "don't upgrade" may be easy to say but it isn't really supported -
> neither at the core, where Tim's quick to point out that we're on our own
> for bug fixes, nor among the external contributors who have no notion that
> any such thing is going on, nor among client sites who are apt to install
> whatever's current.  Is this fun?  Doesn't sound like it.  Will it be more
> fun to artificially prolong the process?  Surely it will not.  Why not
> just do what you have to do, and concede that a few releases hence will
> be bleeding edge, not for general production use?

You have a point, although a problem is that it's not like I know exactly
where I want the language to end up -- I'm continually updating the
direction
based upon feedback.  I don't want to end up with the second system
syndrome,
where everything gets rewritten and redesigned from scratch, but the result
is so different that it takes years to get the user community to migrate.

We're discussing this in python-dev right now, I'd welcome your views there
(note that python-dev no longer has a closed-subscription policy).

--Guido van Rossum (home page: http://www.python.org/~guido/)