max(a,b) sometimes depends on order of args

[alex77g2]

Bug report

Bug description:

# Python 3.11.8 (tags/v3.11.8:db85d51, Feb  6 2024, 22:03:32) [MSC v.1937 64 bit (AMD64)] on win32
>>> max(3.5, float('nan'))
3.5
>>> max(float('nan'), 3.5)
nan
>>> f = float('nan')
>>> f == f
False # NAN per definition does equal itself, so this is the test for nan without functions. Same for other compares.

As this is not obvious behaviour (and python wants to be easy to understand), I suggest to mention this here:
https://docs.python.org/3/library/functions.html#max
I'm not sure about considering this a bug, but is is on the edge between bug and confusing behaviour.

CPython versions tested on:

3.11

Operating systems tested on:

Linux, Windows

[Eclips4]

NAN per definition does equal itself, so this is the test for nan without functions. Same for other compares.

IEEE-754 says the opposite:

Any comparison with a NaN is treated as unordered.

[hugovk]

Looks like a duplicate of #88536.

If math.nan is the first argument for either max() or min(), the result is always nan, regardless of the other values in the results. However, if nan is in any other position in the arguments list, the result is always what you would expect if nan was not in the arguments list.
...

[alex77g2]

yes, but this is not obvious to each python user. and python is intended to be as friendly or comfortable or easy to understand as possible. No hint in documentation?

[Eclips4]

yes, but this is not obvious to each python user. and python is intended to be as friendly or comfortable or easy to understand as possible. No hint in documentation?

There's already a hint in documentation: https://docs.python.org/3/library/math.html#math.nan:

Due to the requirements of the IEEE-754 standard, math.nan and float('nan') are not considered to equal to any other numeric value, including themselves.

[alex77g2]

I mean a hint in max(), min() docu. As not everybody ready everything. I assume many users dont even know about NAN.

[Eclips4]

I mean a hint in max(), min() docu. As not everybody ready everything. I assume many users dont even know about NAN.

This behaviour currently is under discussion, please read the issue found by Hugo.