New test failure
opoplawski opened this issue · comments
Starting to get build failures - https://apps.fedoraproject.org/koschei/package/netcdf4-python?collection=f27
Failure is:
+ PYTHONPATH=../build/lib.linux-x86_64-2.7
+ /usr/bin/python2 run_all.py
...............................................F............E.............................
======================================================================
ERROR: runTest (tst_enum.EnumTestCase)
testing enum data type
----------------------------------------------------------------------
Traceback (most recent call last):
File "/builddir/build/BUILD/netcdf4-python-1.2.7rel/test/tst_enum.py", line 36, in setUp
cloud_var[:] = datain_masked
File "netCDF4/_netCDF4.pyx", line 4059, in netCDF4._netCDF4.Variable.__setitem__ (netCDF4/_netCDF4.c:46083)
ValueError: trying to assign illegal value to Enum variable
======================================================================
FAIL: runTest (tst_masked.PrimitiveTypesTestCase)
testing auto-conversion of masked arrays and packed integers
----------------------------------------------------------------------
not running tst_unicode3.py ...
netcdf4-python version: 1.2.7
HDF5 lib version: 1.8.18
netcdf lib version: 4.4.1.1
numpy version 1.13.0rc1
Traceback (most recent call last):
File "/builddir/build/BUILD/netcdf4-python-1.2.7rel/test/tst_masked.py", line 111, in runTest
assert(file['v'][0] is NP.ma.masked)
AssertionError
----------------------------------------------------------------------
Ran 90 tests in 31.707s
FAILED (failures=1, errors=1)
Not much change from the previous successful build so not sure what is up. What other information would be useful?
I'm guessing its numpy related. Do the tests pass with numpy 1.12.1?
I confirmed that these failures are related to changes in the way masked arrays are handled in numpy 1.13. Pull request coming soon.
Looks to me like this is due to a regression in numpy 1.13.rc0. I've reported this in numpy/numpy#9121.
Fix for failing tst_enum.py in pull request #663 - tst_masked.py still fails because of change in behaviour of scalar numpy masked arrays in 1.13.
OK, received clarification from the numpy folks that this change was intentional and implemented their suggested fix in pull request #663. Tests pass now with numpy 1.13.0rc1.
merging pull request now...