Maintenance/1.10.x #7354
Maintenance/1.10.x #7354
Conversation
Backport of gh-6211 - fix Linux build with Intel compilers
Backport compiler fixes
…raising an ImportError.
npyiter can change its inner stride when it is buffered, this happens in some cases (the test is one of these) mostly for weirdly strided arrays. The test does depend on the buffer size being used. closes gh-6221
It may be that you can say that it cannot happen when grow inner is not enabled as well.
Index iter bug
BUG: simplify code remove strict aliasing violation
BUG: Fix refcounting for string comparison in array_richcompare
Timedelta multiplication and division relied on undefined behaviour for some inputs, which would give the expected results on x86 but not on armv7l (e.g. Raspberry Pi 2). Closes #6274.
Backport-gh-6275 BUG: fix timedelta arithmetic with invalid values or NaTs
REL: Reset strides for RELAXED_STRIDE_CHECKING for 1.10 release.
Issue: _exec_command function doesn't close 'so_dup' and 'se_dup' file descriptors.SciPy try to build scipy\special\amos\zunik.f and crash:error: Command "C:\Program Files (x86)\Intel\Composer XE 2015\bin\intel64\ifort.exe /nologo /MD /nbs /names:lowercase /assume:underscore /O1 -IC:\Python27\lib\site-packages\numpy\core\include -c /c scipy\special\amos\zunik.f /Fobuild\temp.win-amd64-2.7\scipy\special\amos\zunik.o" failed with exit status -1073741502
behavior affect at least python 3.5 and SciPy build and build failed.
During initialization <python>.distutils.MSVCCompiler replace Intel
environment('include' and 'lib' paths). This fix decorate 'initialize'
function in MSVCCompiler and extend 'lib' and 'include' environment
variables. Changed compilation keys: generate optimized code
specialized for Intel processors with SSE4.2 support.
folder with whitespaces.
distutils.msvccompiler.MSVCCompiler
The extra_link_args is sadly not intrinsically used for many parts of the system_info code. This commit adds the linking properties stored when using extra_link_args in the openblas section to bypass any difficulties in the usage of OpenBLAS. This is especially helpful when linking against external LAPACK libraries which requires -lgfortran and possibly -lm for correct linking.
The pavement script was not using the new '.dev0+' GIT_REVISION[:7] suffix for development releases.
Those files in numpy/core/src/multiarray were only included in source distributions when HAVE_CBLAS was defined, making the distribution contents depend on the local configuration. Closes #6343.
Added tests for the x.x.dev0+1234567 form together with alpha and beta versions of same. Updates lifted from scipy.
Fix sdist missing files
This was buggy, because the underlying functions supported it partially but cond was not aware of this. Closes gh-6351
BUG,ENH: allow linalg.cond to work on a stack of matrices
DOC,BUG: Fix some latex generation problems.
Backport 6942
Backport 6946
Backport 6950
Backport 6954
Backport 6952
[ci skip]
Backport 6958
There were problems with the PyPi upload, so make a 1.10.4 release. The new 1.10.4-notes are copies of the old 1.10.3-notes, and the latter have been marked void. [ci skip]
DOC: Prepare for 1.10.4 release.
DOC: Fix typo.
MAINT: Constrain `pip` and `setuptools` versions (Python 3.2 support) (NumPy 1.10.x Backport)
Traceback (most recent call last):
File "C:/Users/Andrei/PycharmProjects/Pharmacosensitivity_growth_assays/src/post_processing.py", line 477, in <module>
all_cell_lines_arr, means_accumulator, errs_accumulator, names_accumulator = stack_data_in_range_of_interest(concs_effective_range)
File "C:/Users/Andrei/PycharmProjects/Pharmacosensitivity_growth_assays/src/post_processing.py", line 129, in stack_data_in_range_of_interest
mode='constant', constant_values=((np.nan, np.nan), (np.nan, np.nan)))
File "C:\Anaconda\lib\site-packages\numpy\lib\arraypad.py", line 1358, in pad
cast_to_int=False)
File "C:\Anaconda\lib\site-packages\numpy\lib\arraypad.py", line 1070, in _normalize_shape
return tuple(tuple(axis) for axis in arr.tolist())
AttributeError: 'tuple' object has no attribute 'tolist'
|
|
Ok, took note about master and branch, will try when I have a bit more time to dig in depth. The behavior of the constant padding in multiple dimensions was never documented, which I guess led the the failure of detecting the change in the behavior. Should I still add it into the docstrings? I assume they are run as unit-tests during the continuous integration, right? Should I also investigate the weird behavior of the |
No, we don't test docstrings. If this is behavior that you were depending on and we broke it, though, then the usual default assumption would be that that's a bug and we should fix it (regardless of whether it was actually documented or tested). If it turns out there was some good reason why it was broken or similar than that might change, but if it's just a matter of coercing array-like inputs into real arrays then it's probably just a bug. And if you want to submit a fix then you should certainly also add a test at the same time, and documentation if that makes sense.
This is sorta-intentional, in that if you insist on casting from float to integer then we map |
|
Closing this pull request due to git being difficult -- feel free to re-open a new one :-) |
Added this example for
padfunction inarraypad.py:with respect to undocumented change in behavior from issue #7353