[Python-Dev] [Python-checkins] r64424 - inpython/trunk:Include/object.h Lib/test/test_sys.pyMisc/NEWSObjects/intobject.c Objects/longobject.cObjects/typeobject.cPython/bltinmodule.c

[Steven D'Aprano]

On Fri, 27 Jun 2008 07:30:43 am Raymond Hettinger wrote:
> The format is already close to the C99 notation
> but replaces the 'p' with '* 2.0 **' which I find to
> be both readable and self-explanatory.

Since we're talking about what's "readable and self-explanatory", I find 
that jarring, unexpected, unintuitive, and mathematically bizarre (even 
if it is the convention in some areas). It's like writing '123 * A.0 
** -2' for 1.23.

And putting spaces around the operators is ugly.

I'd like to mention that what bin() et al is actually doing is not so 
much returning a binary number string but returning a hybrid 
binary/decimal arithmetic expression. So bin() returns a binary number 
string for int arguments, and an expression for float arguments: these 
are conceptually different kinds of things, even if they're both 
strings.

Frankly, I'd be much happier if the API (whatever it is) returned a 
tuple of (binary string, base int, exponent int), and let users write 
their own helper function to format it any way they like. Or failing 
that, the p notation used by Java and C99. (And yes, mixing decimal 
exponents with binary mantissas upsets me too, but somehow it's less 
upsetting.)


-- 
Steven