lgamma(-n) for n = 0, 1, 2, .. does not raise FE_DIVBYZERO.

Question

lgamma(-n) for n = 0, 1, 2, .. does not raise FE_DIVBYZERO.

kargl opened this issue 5 years ago · comments

Tests with olibm:
lgammaf( 1) = 0. Got no exception. Expected 0 without exception.
lgammaf( 2) = 0. Got no exception. Expected 0 without exception.
lgammaf( 0) = inf. Got no exception. Expected inf with FE_DIVBYZERO.
lgammaf(-0) = inf. Got no exception. Expected inf with FE_DIVBYZERO.
lgammaf(-1) = inf. Got no exception. Expected inf with FE_DIVBYZERO.

Tests with FreeBSD libm.
lgammaf( 1) = 0. Got no exception. Expected 0 without exception.
lgammaf( 2) = 0. Got no exception. Expected 0 without exception.
lgammaf( 0) = inf. Got FE_DIVBYZERO. Expected inf with FE_DIVBYZERO.
lgammaf(-0) = inf. Got FE_DIVBYZERO. Expected inf with FE_DIVBYZERO.
lgammaf(-1) = inf. Got FE_DIVBYZERO. Expected inf with FE_DIVBYZERO.

kargl · Answer 1 · Sat Mar 14 2020 06:38:46 GMT+0800 (China Standard Time)

Note, lgammal(-n) correctly raises the FE_DIVBYZERO exception. lgammaf and lgamma have issues.

Oscar Smith · Answer 2 · Tue May 31 2022 23:17:44 GMT+0800 (China Standard Time)

This hasn't yet been fixed