Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

guvectorize does not implement the numpy ufunc interface in versions 0.53+

See original GitHub issue

Hi, after upgrading numba from version 0.52.0 to either versions 0.53.0 or 0.54.0, I have found that the signature field of functions compiled using @guvectorize no longer is implemented.

Tested with: numpy version 1.20.0 numba version 0.52.0, 0.53.0, 0.54.0

Here is a minimal working example:

from numba import guvectorize

@guvectorize(["void(float64[:], float64[:], float64[:])"],
             "(n),(n)->()", nopython=True, cache=True)
def test(a, b, c):
    c[0]=0
    for i in range(len(a)):
        c[0] += a[i]*b[i]

When I try using this, here’s what happens with 0.52.0:

In [1]: import test_gufunc                                                      

In [2]: test_gufunc.test.signature                                              
Out[2]: '(n),(n)->()'

And here’s what I get with the newer versions:

In [1]: import test_gufunc                                                      

In [2]: test_gufunc.test.signature                                              
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-2-d4d284438f75> in <module>
----> 1 test_gufunc.test.signature

AttributeError: 'GUFunc' object has no attribute 'signature'

Reporting a bug

I have tried using the latest released version of Numba (most recent is visible in the change log (https://github.com/numba/numba/blob/master/CHANGE_LOG).
I have included a self contained code sample to reproduce the problem. i.e. it’s possible to run as ‘python bug.py’.