[Python-Dev] Rich comparisons
Michael Hudson
mwh at python.net
Tue Mar 30 05:49:42 EST 2004
"Tim Peters" <tim.one at comcast.net> writes:
[>>> is Ed Loper, >> is me]
>>> I'm also disturbed by the fact that cmp() has something different to
>>> say about their equality:
>>>
>>> >>> cmp(float('nan'), float('nan'))
>>> 0
>
>> Well, yah. cmp() assumes a total ordering. If there just isn't one,
>> what can we do?
>
> If we could recognize it was a NaN, my preference would be to raise an
> exception.
Well, we could sorta do this, by doing something like
if a > b: return 1
if a < b: return -1
if a == b: return 0
raise ValueError, 'can't compare NaNs'
which again goes back to depending on the C compiler.
> I was told that GNU sort arranges to make NaNs look "less than" all
> non-NaN floats, which is a way to make a total ordering -- kinda.
Well, it's an answer. Not sure it's a good one.
>> I have at no point claimed that I have given Python 2.4 a coherent
>> ieee 754 floating point story.
>
> Ya, but I keep telling people you claim that <wink>.
Oh! So it's your fault <wink>.
>> If someone (not me!) has the time and energy to do a 'proper job' (and
>> I'd expect working out what that means to be the hard part), then you
>> have my support and pre-emptive thanks.
>
> I don't believe it will happen until all the major C compilers support the
> (still optional!) 754 gimmicks in C99 -- or Python is rewritten in some
> other language that supports them. Before then, we'd need a small army of
> people expert in the intersection of their platform C and the 754 standard,
And Python's implementation!
> to produce piles of #ifdef'ed platform-specific implementation code.
> That's doable, but unlikely.
T'would be easier in assembly (as would checking for overflow in int
multiplication, but that's another story).
I could do 'it' in PPC assembly for MacOS X if I knew what 'it' was,
but I'm not a hardcore numerics geek, I was just offended by
float('nan') == 1 returning True.
Cheers,
mwh
--
Haha! You had a *really* weak argument! <wink>
-- Moshe Zadka, comp.lang.python
More information about the Python-Dev
mailing list