summaryrefslogtreecommitdiff
path: root/numpy
Commit message (Collapse)AuthorAgeFilesLines
* DOC: Fix obsolete data in readmefix-8381-docsEric Wieser2017-02-211-6/+2
| | | | Link was dead, and this project is neither Numeric nor in CVS any more. This should have gone into #8381
* Merge pull request #8381 from eric-wieser/rebuild-lapack_liteCharles Harris2017-02-204-7794/+2972
|\ | | | | MAINT: Rebuild lapack lite
| * BUG: add missing routines to the transpilation listEric Wieser2016-12-294-553/+72104
| | | | | | | | These are taken from lapack_litemodule.c and umath_linalg.c.src
| * MAINT: Sort the functions, for ease of diffingEric Wieser2016-12-291-10/+10
| |
| * BUG: Do not repeat xerbla_, which we deliberately replaceEric Wieser2016-12-292-64/+8
| | | | | | | | This is also defined in python_xerbla.c, where we redefine a python-compatible version
| * MAINT: retranspile lapack from the supposed sourceEric Wieser2016-12-293-78001/+1684
| | | | | | | | The source used is http://archive.debian.org/debian/pool/main/l/lapack3/lapack3_3.0.20000531a.orig.tar.gz. Originally this was done with a patched f2c, but if the patch isn't provided in the source tree, there's no sensible way to use it
* | Merge pull request #8594 from eric-wieser/MaskedArray.__setitem__Marten van Kerkwijk2017-02-202-1/+8
|\ \ | | | | | | BUG: Fix MaskedArray.__setitem__
| * | BUG: Fix #8510, making MaskedArray.__setitem__ workEric Wieser2017-02-202-1/+8
| | |
* | | Merge pull request #8638 from juliantaylor/packbits-fixCharles Harris2017-02-192-1/+10
|\ \ \ | | | | | | | | BUG: fix wrong odd determination in packbits
| * | | BUG: fix wrong odd determination in packbitsJulian Taylor2017-02-192-1/+10
| | | | | | | | | | | | | | | | closes gh-8637
* | | | Merge pull request #8645 from eric-wieser/ma-use-getmaskMarten van Kerkwijk2017-02-191-16/+15
|\ \ \ \ | | | | | | | | | | MAINT: Use getmask where possible
| * | | | MAINT: Use getmaskarrayEric Wieser2017-02-091-2/+1
| | | | | | | | | | | | | | | | | | | | | | | | | This is subtly different in behaviour - before, if obj._mask == no_mask, the array was not created. However, this case is not allowed anyway, so this doesn't matter
| * | | | MAINT: Use getmask where possibleEric Wieser2017-02-091-14/+14
| | | | |
* | | | | Merge pull request #8618 from eric-wieser/fix-8507Marten van Kerkwijk2017-02-192-3/+14
|\ \ \ \ \ | | | | | | | | | | | | BUG: Fix undefined behaviour induced by bad __array_wrap__
| * | | | | BUG: Fixed undefined behavior when __array_wrap__ returns NoneEric Wieser2017-02-192-3/+14
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Now `None` is treated like any other return value, rather than a command to segfault.
* | | | | | Merge pull request #8629 from eric-wieser/speedup-indicesMarten van Kerkwijk2017-02-191-7/+4
|\ \ \ \ \ \ | | | | | | | | | | | | | | ENH: Improve the efficiency of indices
| * | | | | | ENH: Improve the efficiency of indicesEric Wieser2017-02-181-7/+4
| |/ / / / / | | | | | | | | | | | | | | | | | | | | | | | | Previously, this special cased something that needed no special case, and did addition with zero to encourage broadcasting that would happen anyway.
* | | | | | Merge pull request #8633 from MSeifert04/boolean_arrays_in_ix__funcEric Wieser2017-02-191-1/+14
|\ \ \ \ \ \ | | | | | | | | | | | | | | DOC: Mention boolean arrays in the ix_ documentation.
| * | | | | | DOC: Mention boolean arrays in the ix_ documentation.Michael Seifert2017-02-181-1/+14
| | | | | | |
* | | | | | | Merge pull request #8631 from anntzer/fill_diagonal-typoEric Wieser2017-02-191-2/+2
|\ \ \ \ \ \ \ | |_|_|_|/ / / |/| | | | | | Fix typo in fill_diagonal docstring.
| * | | | | | DOC: Fix typo in fill_diagonal docstring.Antony Lee2017-02-181-2/+2
| |/ / / / /
* | | | | | Merge pull request #8614 from eric-wieser/apply_along_axis-emptyMarten van Kerkwijk2017-02-182-1/+23
|\ \ \ \ \ \ | |/ / / / / |/| | | | | BUG: Don't leak internal exceptions when given an empty array
| * | | | | BUG: Don't leak internal exceptions when given an empty arrayEric Wieser2017-02-132-1/+23
| |/ / / / | | | | | | | | | | | | | | | Fixes #7454
* | | | | Merge pull request #8043 from pv/ufunc-copy-overlapCharles Harris2017-02-1616-126/+1035
|\ \ \ \ \ | | | | | | | | | | | | ENH: umath: ensure ufuncs are well-defined with memory overlapping inputs
| * | | | | DOC: core: remove old comment saying public API may changePauli Virtanen2017-02-101-5/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | It's unlikely we will break backward compatibility, so better not claim we will do it in the code.
| * | | | | STY: core: fix up code style and add clearer commentsPauli Virtanen2017-02-104-18/+38
| | | | | |
| * | | | | BUG: core: fix refcount errors on error handlingPauli Virtanen2017-02-063-1/+2
| | | | | |
| * | | | | BUG: multiarray: fix refcount error in error handling in ↵Pauli Virtanen2017-01-241-1/+0
| | | | | | | | | | | | | | | | | | | | | | | | PyArray_MapIterArrayCopyIfOverlap
| * | | | | DOC: fix wrong flag name docstringPauli Virtanen2017-01-231-1/+1
| | | | | |
| * | | | | DOC: core/umath: make a comment in the test be less misleadingPauli Virtanen2017-01-221-1/+3
| | | | | |
| * | | | | BUG: core/umath: handle __array_prepare__ correctly in ufunc iterator_loopPauli Virtanen2017-01-212-17/+53
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Ensure that results get written to the array returned by __array_prepare__, if any. This necessitates calling prepare_ufunc_output on the output arrays that the iterator may replace with UPDATEIFCOPY'd ones. Add a test to the test suite --- this code path was untested previously.
| * | | | | BUG: core/umath: fix cornercase for overlapping operands with internal overlapPauli Virtanen2017-01-213-21/+56
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Ensure that a copy is made when ufunc output operand overlaps with ufunc input, and both have internal overlap. The heuristic applied when iteration order is elementwise is OK only in the absence of internal overlap.
| * | | | | ENH: core: rename NPY_ITER_OVERLAP_ALLOW_SAME to ↵Pauli Virtanen2017-01-215-15/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | NPY_ITER_OVERLAP_ASSUME_ELEMENTWISE Make the flag name indicate more clearly on what is the precise assumption made.
| * | | | | TST: core: streamline and extend test_mem_overlap.TestUFunc testsPauli Virtanen2017-01-201-60/+54
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Use itertools to go over all permutations rather than trying to do it manually.
| * | | | | BUG: core: fix bug with zero strides in PyArray_EQUIVALENTLY_ITERABLE_OVERLAP_OKPauli Virtanen2017-01-201-4/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Arrays with zero strides are never "ahead", as the same element is reused multiple times. The previous logic was incorrect eg. for stride1==0, stride2<0 and arr1->data==arr2->data.
| * | | | | MAINT: core: fix compiler warningsPauli Virtanen2017-01-202-3/+1
| | | | | |
| * | | | | DOC: fix typo in commentPauli Virtanen2017-01-191-1/+1
| | | | | |
| * | | | | STY: break long linePauli Virtanen2017-01-191-2/+2
| | | | | |
| * | | | | BUG: umath: correct treatment of __array_prepare__Pauli Virtanen2017-01-191-3/+11
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | prepare_ufunc_output is allowed to replace the array, because we check that it's still a view. Also if COPY_IF_OVERLAP made updateifcopy of op[i], the data will be reflected in the array from __array_prepare__. The is-a-view check is only bare-bones, but it was so since 61d45eee
| * | | | | MAINT: drop NPY_INLINEPauli Virtanen2017-01-191-1/+1
| | | | | |
| * | | | | MAINT: bump API versionPauli Virtanen2017-01-193-2/+6
| | | | | |
| * | | | | ENH: core: change NPY_ITER_OVERLAP_NOT_SAME to NPY_ITER_OVERLAP_ALLOW_SAMEPauli Virtanen2017-01-196-22/+34
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | It's clearer to explicitly specify which operands are accessed elementwise, rather than to list exceptions.
| * | | | | STY: core: remove unused variablePauli Virtanen2017-01-191-1/+0
| | | | | |
| * | | | | BUG: umath: in reduceat, arrays must be copied on overlap even if they are ↵Pauli Virtanen2017-01-194-2/+39
| | | | | | | | | | | | | | | | | | | | | | | | the same
| * | | | | ENH: core: handle memory overlap in ufunc.atPauli Virtanen2017-01-194-11/+120
| | | | | | | | | | | | | | | | | | | | | | | | This adds a new method PyArray_MapIterArrayCopyIfOverlap to the API.
| * | | | | TST: core: add tests for memory overlap in ufunc reductionsPauli Virtanen2017-01-191-15/+114
| | | | | |
| * | | | | ENH: umath: turn on overlap detection + copying in ufunc reductionsPauli Virtanen2017-01-192-6/+51
| | | | | |
| * | | | | TST: core: add tests to check that basic ufunc calls deal with memory overlapPauli Virtanen2017-01-191-22/+274
| | | | | |
| * | | | | BUG: core/ufunc: fix assumptions about iterator in execute_fancy_ufunc_loopsPauli Virtanen2017-01-191-21/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Because the wheremask loop does not write to all elements of the array, it needs to consider output arrays as READWRITE, in case COPY_IF_OVERLAP makes temporary copies. As the iterator may make temporary copies of arrays, make sure to pass those to __array_prepare__ instead of the originals.
| * | | | | ENH: umath: turn on overlap detection + copying in ufunc basic callsPauli Virtanen2017-01-191-3/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | After this commit, all code paths in PyUFunc_GeneralizedFunction and PyUFunc_GenericFunction perform input/output operand memory overlap detection, and make copies to eliminate it.
View Lorry Controller Status