Proposal: min(None, x) and max(None, x) return x

[Terry Reedy]

"Eric Brunel" <eric.brunel at pragmadev.com> wrote in message
news:arle6m$aa9$1 at news-reader11.wanadoo.fr...
> Hi all,
>
> I quite often find myself doing things like:
>
> xMax = None
> for -whatever-:
>   -big calculation leading to a value of x-
>   if xMax is None:
>     xMax = x
>   else:
>     xMax = max(x, xMax)
>
> or the same kind of stuff with min, of course.
>
> So I told myself: wouldn't it be great if max(None, x) or min(None,
x)
> always simply returned x? So I could just write:
>
> xMax = None
> for -whatever-:
>   -big calculation leading to a value of x-
>   xMax = max(x, xMax)
>
> Okay, it only saves 3 lines, but I personally find it quite
natural... In
> addition, today, min(None, x) and max(None, x) work, but have
strange
> results: as far as I can see, min(None, x) is always None and
max(None, x)
> is always x, but this behaviour isn't documented.
>
> Opinions anyone?

What you want are identity elements for the min and max operators
(like 0,1 are for +,*), which are -+ infinity.  In a particular
context, you may be able to use finite values for the same purpose.
On some builds (depending on underlying C library), you can actually
get and use values representing and acting like +-infinity.  On AS 2.2
Windows build:

>>> Infplus=1e300/1e-300
>>> Infplus
1.#INF
>>> max(1e300,Infplus)
1.#INF
>>> max(-1e300,Infneg)
-1.0000000000000001e+300

>>> Infneg=-1e300/1e-300
>>> Infneg
-1.#INF
>>> min(-1e300,Infneg)
-1.#INF
>>> max(-1e300,Infneg)
-1.0000000000000001e+300

So proposal unnecessary on this build.  What you say in last para is
that None is treated as -INFINITY, which is both arbitrary and a bit
weird.

Terry J. Reedy