float / double support in Python?

[Brian Quinlan]

> > Supporting a basic machine type (in a zero-overhead fashion) would
> > require an amazing amount of effort for a language like Python.
> 
> It doesn't have to be zero overhead.  It needs to be less overhead
than
> doubles, both in space and time.

Are you saying that a Python single needs to have less overhead than a
Python double or that a Python single needs to have less overhead than a
C double? 

The former makes no sense unless it is acceptable for a Python single to
be 1000x less efficient than a C double so long as a Python double is
1001x less efficient than a C double.

The later is difficult, as I said.

> > That
> > effort would be better directed to features that more of the
community
> > is interested in.
> 
> If the community retains a provincial interest, rather than noticing
what
> the language is actually getting used for and what certain people
want,
> then the community will not grow.  It will not receive mainstream 
> attention, and it will become a "coulda woulda shoulda" language. 

In order of interest by the majority of programmers, how would you rank
these:

- robust date/time handling
- robust database access libraries
- robust XML libraries
- support for fixed-point type
- support for logging facilities
- robust i18n support
- support for C singles

Now guess which of those is NOT going to appear/be improved in Python
2.3? I also think that there is a significant diversity of work being
done that you can't reasonably call the community "provincial".

Cheers,
Brian