merge master into 1.9 branch #4849
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…laboration of the problems with mixing np.ndarray and np.matrix
…ation, so couch it more cautiously
MAINT: const correctness and minor fixes to C code
FIX pickling ufuncs defined in nested modules
clongdouble type required alignment of 32 bytes which is never provided by numpys allocation scheme so they always took unaligned paths. As complex types are structs of two smaller types only half the alignment is actually required for operations on them. Copying complex types on the other hand might require a larger alignment as copies can be done on the whole structure so one can not just remove the doubling of the alignment. As a compromise limit the maximum alignment to 16 bytes which is double what the current numpy copy loops require and does not penalize amd64 with unnecessary unaligned code paths in most cases. 16 bytes might be required on sparc to load long doubles. Complex doubles on i386 will still unnecessarly take unaligned code paths as the system will not provide 16 byte alignment. Closes numpygh-3768
e.g. ufunc_at wraps input arrays which without this fix causes the iterator to choose much slower unaligned code paths. Also pointers from malloc can be not aligned for long double types, e.g. on debian sparc.
transposing structured arrays can lose alignment
(debian) sparc system malloc does not provide the alignment required by 16 byte long double types this means the inout intent cannot be satisfied and several tests fail as the alignment flag can be randomly true or fals when numpy gains an aligned allocator the tests could be enabled again.
when orders are selected where the kth element falls into an equal range the the last stored pivot was not the kth element, this leads to losing the ordering of smaller orders as following selection steps can start at index 0 again instead of the at the offset of the last selection. Closes numpygh-4836
BUG: wrong selection for orders falling into equal ranges
the number types does not match the size of the blank array anymore, use a larger array and add asserts to ensure the size matches.
Uses old code to be more broad then previous commit
When a cython object contains numpy arrays the pure python allocation hook can trigger during cythons __dealloc__ method and trigger a second deletion of the object currently being deleted. To minimize the probabily that that happens disable the garbage collector during the hook. As this involves python calls it is still possible that a double delete occurs but chances are lowered, a proper solution would be C only hook like python 3.4 tracemalloc module. Closes numpygh-4834
BUG: disable garbage collector during memory allocation hook
WIP: Fix matplotlib, etc. errors
BUG: retain writeable flag when indexing subclasses
Fix spelling and grammar in numpy/core/src/multiarray/mapping.c numpy/core/tests/test_indexing.py numpy/core/tests/test_regression.py
BUG: fix buffer overflow in data array of ldexp and frexp
MAINT: Spellcheck some files.
A new PEP for infix matrix multiplication
better writeable flag fix
DOC: Update 1.9.0-notes to mention pairwise summation.
juliantaylor
added a commit
that referenced
this pull request
Jul 7, 2014
merge master into 1.9 branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
as we have not opened 1.10 yet we can merge it into 1.9 to preserve ancestry of the changes and simplify backports