Jim Bosch added a comment - 08/Jun/18 9:35 AM I'm quite certain that we're not relying on any extra attributes being copied.  I'd guess NumPy has a way to squash the warning by acknowledging somehow (i.e. with some keyword argument) that you accept the changed behavior.

Tim Jenness added a comment - 08/Mar/22 5:45 PM The code triggering this is from polynomial.py: 1325 def __mul__(self, other): 1326 if isscalar(other): 1327 return poly1d(self.coeffs * other) 1328 else: 1329 other = poly1d(other) 1330 return poly1d(polymul(self.coeffs, other.coeffs)) 1331 The warning from polynomial.py comes from: 1223 if set(c_or_r.__dict__) - set(self.__dict__): 1224 msg = ("In the future extra properties will not be copied " 1225 "across when constructing one poly1d from another") 1226 warnings.warn(msg, FutureWarning, stacklevel=2) 1227 self.__dict__.update(c_or_r.__dict__) The full stack trace is: Traceback (most recent call last): File "tests/test_hermiteTransformMatrix.py", line 83, in testTransformMatrix v1 = self.ht(inx)(transPoint.getX()) * self.ht(iny)(transPoint.getY()) File "tests/test_hermiteTransformMatrix.py", line 49, in ht return (scipy.poly1d([(2**n * np.pi**0.5 * scipy.special.gamma(n+1))**(-0.5)]) File "/Users/timj/work/lsstsw3/miniconda/envs/lsst-scipipe-2.0.0/lib/python3.8/site-packages/numpy/lib/polynomial.py", line 1329, in __mul__ other = poly1d(other) File "/Users/timj/work/lsstsw3/miniconda/envs/lsst-scipipe-2.0.0/lib/python3.8/site-packages/numpy/lib/polynomial.py", line 1226, in __init__ warnings.warn(msg, FutureWarning, stacklevel=2)

Jim Bosch added a comment - 09/Mar/22 4:54 PM Still seems valid.  Low priority because pretty much nobody develops in this package anyway, so these warnings are much less visible than others.