BUG: implementation of npy_get_floatstatus_barrier incomplete for musl libc

I’ve ran into issue with linalg.solve on different systems: in RHEL and Ubuntu it will raise an LinAlgError exception, but on Alpine it will return nan.

Ubuntu:

$ docker run -it --rm ubuntu:18.04 /bin/sh
$ apt-get install python3-numpy
# python3
Python 3.6.6 (default, Sep 12 2018, 18:26:19)
[GCC 8.0.1 20180414 (experimental) [trunk revision 259383]] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy
>>> numpy.linalg.solve(numpy.array([[0]]), numpy.array([[0]]))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3/dist-packages/numpy/linalg/linalg.py", line 375, in solve
    r = gufunc(a, b, signature=signature, extobj=extobj)
  File "/usr/lib/python3/dist-packages/numpy/linalg/linalg.py", line 90, in _raise_linalgerror_singular
    raise LinAlgError("Singular matrix")
numpy.linalg.linalg.LinAlgError: Singular matrix

Alpine:

$ docker run -it --rm alpine:3.8 /bin/sh
# apk --update add python3 py3-numpy
/ # python3
Python 3.6.6 (default, Aug 24 2018, 05:04:18)
[GCC 6.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy
>>> numpy.linalg.solve(numpy.array([[0]]), numpy.array([[0]]))
array([[nan]])

I would expect to receive same exception on Alpine as well. Is this known issue?

I’ve tried to use either lapack or openblas on alpine, compile numpy from source — it’s still would return nan.

[mattip]

ok, thanks. @sashkab see this comment for why that happens.

[sashkab]

@sashkab Thanks. What was the unused function error when you had '-Werror=unused-function' defined?

They were:

In file included from numpy/core/src/umath/loops.c.src:47:0:
numpy/core/src/umath/simd.inc.src:1020:1: error: 'sse2_minimum_DOUBLE' defined but not used [-Werror=unused-function]
 sse2_@kind@_@TYPE@(@type@ * ip, @type@ * op, const npy_intp n)
 ^~~~~~~~~~~~~~~~~~~
In file included from numpy/core/src/umath/loops.c.src:47:0:
numpy/core/src/umath/simd.inc.src:1020:1: error: 'sse2_maximum_DOUBLE' defined but not used [-Werror=unused-function]
 sse2_@kind@_@TYPE@(@type@ * ip, @type@ * op, const npy_intp n)
 ^~~~~~~~~~~~~~~~~~~
In file included from numpy/core/src/umath/loops.c.src:47:0:
numpy/core/src/umath/simd.inc.src:1020:1: error: 'sse2_minimum_FLOAT' defined but not used [-Werror=unused-function]
 sse2_@kind@_@TYPE@(@type@ * ip, @type@ * op, const npy_intp n)
 ^~~~~~~~~~~~~~~~~~
In file included from numpy/core/src/umath/loops.c.src:47:0:
numpy/core/src/umath/simd.inc.src:1020:1: error: 'sse2_maximum_FLOAT' defined but not used [-Werror=unused-function]
 sse2_@kind@_@TYPE@(@type@ * ip, @type@ * op, const npy_intp n)
 ^~~~~~~~~~~~~~~~~~
cc1: some warnings being treated as errors
In file included from numpy/core/src/umath/loops.c.src:47:0:
numpy/core/src/umath/simd.inc.src:1020:1: error: 'sse2_minimum_DOUBLE' defined but not used [-Werror=unused-function]
 sse2_@kind@_@TYPE@(@type@ * ip, @type@ * op, const npy_intp n)
 ^~~~~~~~~~~~~~~~~~~
In file included from numpy/core/src/umath/loops.c.src:47:0:
numpy/core/src/umath/simd.inc.src:1020:1: error: 'sse2_maximum_DOUBLE' defined but not used [-Werror=unused-function]
 sse2_@kind@_@TYPE@(@type@ * ip, @type@ * op, const npy_intp n)
 ^~~~~~~~~~~~~~~~~~~
In file included from numpy/core/src/umath/loops.c.src:47:0:
numpy/core/src/umath/simd.inc.src:1020:1: error: 'sse2_minimum_FLOAT' defined but not used [-Werror=unused-function]
 sse2_@kind@_@TYPE@(@type@ * ip, @type@ * op, const npy_intp n)
 ^~~~~~~~~~~~~~~~~~
In file included from numpy/core/src/umath/loops.c.src:47:0:
numpy/core/src/umath/simd.inc.src:1020:1: error: 'sse2_maximum_FLOAT' defined but not used [-Werror=unused-function]
 sse2_@kind@_@TYPE@(@type@ * ip, @type@ * op, const npy_intp n)
 ^~~~~~~~~~~~~~~~~~
cc1: some warnings being treated as errors