NumPy 1.7.2 Release Notes

This is a bugfix only release in the 1.7.x series. It supports Python 2.4 - 2.7 and 3.1 - 3.3 and is the last series that supports Python 2.4 - 2.5.

Issues fixed

  • gh-3153: Do not reuse nditer buffers when not filled enough

  • gh-3192: f2py crashes with UnboundLocalError exception

  • gh-442: Concatenate with axis=None now requires equal number of array elements

  • gh-2485: Fix for astype(‘S’) string truncate issue

  • gh-3312: bug in count_nonzero

  • gh-2684: numpy.ma.average casts complex to float under certain conditions

  • gh-2403: masked array with named components does not behave as expected

  • gh-2495: np.ma.compress treated inputs in wrong order

  • gh-576: add __len__ method to ma.mvoid

  • gh-3364: reduce performance regression of mmap slicing

  • gh-3421: fix non-swapping strided copies in GetStridedCopySwap

  • gh-3373: fix small leak in datetime metadata initialization

  • gh-2791: add platform specific python include directories to search paths

  • gh-3168: fix undefined function and add integer divisions

  • gh-3301: memmap does not work with TemporaryFile in python3

  • gh-3057: distutils.misc_util.get_shared_lib_extension returns wrong debug extension

  • gh-3472: add module extensions to load_library search list

  • gh-3324: Make comparison function (gt, ge, …) respect __array_priority__

  • gh-3497: np.insert behaves incorrectly with argument ‘axis=-1’

  • gh-3541: make preprocessor tests consistent in halffloat.c

  • gh-3458: array_ass_boolean_subscript() writes ‘non-existent’ data to array

  • gh-2892: Regression in ufunc.reduceat with zero-sized index array

  • gh-3608: Regression when filling struct from tuple

  • gh-3701: add support for Python 3.4 ast.NameConstant

  • gh-3712: do not assume that GIL is enabled in xerbla

  • gh-3712: fix LAPACK error handling in lapack_litemodule

  • gh-3728: f2py fix decref on wrong object

  • gh-3743: Hash changed signature in Python 3.3

  • gh-3793: scalar int hashing broken on 64 bit python3

  • gh-3160: SandboxViolation easyinstalling 1.7.0 on Mac OS X 10.8.3

  • gh-3871: npy_math.h has invalid isinf for Solaris with SUNWspro12.2

  • gh-2561: Disable check for oldstyle classes in python3

  • gh-3900: Ensure NotImplemented is passed on in MaskedArray ufunc’s

  • gh-2052: del scalar subscript causes segfault

  • gh-3832: fix a few uninitialized uses and memleaks

  • gh-3971: f2py changed string.lowercase to string.ascii_lowercase for python3

  • gh-3480: numpy.random.binomial raised ValueError for n == 0

  • gh-3992: hypot(inf, 0) shouldn’t raise a warning, hypot(inf, inf) wrong result

  • gh-4018: Segmentation fault dealing with very large arrays

  • gh-4094: fix NaT handling in _strided_to_strided_string_to_datetime

  • gh-4051: fix uninitialized use in _strided_to_strided_string_to_datetime

  • gh-4123: lexsort segfault

  • gh-4141: Fix a few issues that show up with python 3.4b1