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

[Paul Boddie]

Guido van Rossum <guido at python.org> wrote in message news:<3CB1CEC0.F324F480 at python.org>...
>
> Paul wrote:
> >
> > As far as the 2.2 features are concerned, they seem to have a
> > reputation in certain circles as being badly-defined, immature and
> > "best steered clear of until later".
> 
> Huh?  I'm not sure where you heard that.  Alex Martelli himself just
> told me that he is using it and considers it stable.  Maybe the people
> who say this themselves deserve a reputation as immature and "best steered
> clear of"?

Perhaps their views were tainted by the need to support releases of
Python lower than 2.2. I'm sure that there are many nice features
which reduce the amount of code (along with the complexity of the
code) substantially in Python 2.2, but it isn't always fair to expect
everyone to upgrade Python every time someone releases a new piece of
software.

Even though I use 2.0 and only have 2.1 to hand because an increasing
number of packages required it over time (most of which I only
evaluated anyway), I couldn't see why I also had to install 2.2 when
it was released because the odd package required it, arguably on an
unnecessarily frivolous basis.

I suppose one argument against a "super upgrade" is that backwards
compatibility is readily abandoned (by package authors), but then,
regardless of the scale of the differences between releases, I would
estimate that there is still a fair amount of coding for compatibility
on a number of the larger Python projects. (Although PyXML's support
for Python 1.5.2 is possibly too heroic to be practical any more.)

Paul