cmath.tanh(±0+infj) and tanh(±0+nanj) should return ±0+nanj

[skirpichev]

Bug report

Bug description:

As per C11 DR#471 (accepted for C17), ctanh (0 + i NaN) and ctanh (0 + i Inf) should return 0 + i NaN (with "invalid" exception in the second case). Currently, real part is nan.

This has corresponding implications for ctan(z), as its errors and special cases are handled as if the operation is implemented by -i*ctanh(i*z).

See glibc patch: https://sourceware.org/git/?p=glibc.git;a=commitdiff;h=d15e83c5f5231d971472b5ffc9219d54056ca0f1

CPython versions tested on:

CPython main branch

Operating systems tested on:

No response

Linked PRs

• gh-122637: fix tanh(±0+infj) and tanh(±0+nanj) to return ±0+nanj #122638

[skirpichev]

PR is ready for review: #122638

See also #121889

[mdickinson]

Closed in #122638, with thanks to @skirpichev. As with #121889, I don't think this is behaviour that should be changed in a bugfix release, so it's not a candidate for backport.

[skirpichev]

Great. Now glibc's libm pass cmath_testcases.txt.

One minor note, now we have slightly misleading comment:

cpython/Modules/cmathmodule.c

Lines 157 to 158 in e6fad7a

	/* First, the C functions that do the real work. Each of the c_*
	functions computes and returns the C99 Annex G recommended result

This (and fix for #121889) is available in C17, not C11C99. Though, as you can see, glibc turns this on unconditionally.

Probably, we should adjust comment.

PS: And now this really closed.