[Python-Dev] math.sqrt(-1) - NaN, ValueError or whatever you want?

[Tim Peters]

[Jack Jansen]
> The answers to my math questions never seem to stick, so I'm afraid
> I'm gonna have to ask this question again (even though I think I did
> the same a couple of years ago, but google doesn't find it).
>
> It turns out that on MacOS math.sqrt(-1) returns "nan", as does
> math.log(-1), and I didn't try any other ones.
>
> This is true for all possible combinations of MacPython 2.3, 2.4a0 or
> apple-Python 2.3, MacPython-OS9 or MacPython-OSX, and Mac OS 9 or Mac
> OS X.
>
> When I tried the same on a linux box (Python 2.4a0) it raised
> ValueError: math domain error, which is what I would have expected.

Take this to heart (it's from the current math module docs):

    Note:  The math module consists mostly of thin wrappers around the
    platform C math library functions.  Behavior in exceptional cases is
    loosely specified by the C standards, and Python inherits much of its
    math-function error-reporting behavior from the platform C
    implementation.  As a result, the specific exceptions raised in error
    cases (and even whether some arguments are considered to be exceptional
    at all) are not defined in any useful cross-platform or cross-release
    way.  For example, whether math.log(0) returns -Inf or raises
    ValueError or OverflowError isn't defined, and in cases where
    math.log(0) raises OverflowError, math.log(0L) may raise ValueError
    instead.

> Why this difference?

Platform C.

> Is it something I should fix?

Good luck <wink>.

> How?

You'll understand the "<wink>" above after you give up.

> Or document it?

I think the above already does.  Else you're trying to document a maze of
moving targets (including, on some platforms, the specific compiler options,
link options, and environment variables the user indulges).