summaryrefslogtreecommitdiff
path: root/numpy
Commit message (Collapse)AuthorAgeFilesLines
* MAINT - update comments related to searchsorted to be consistent with code.Bago Amirbekian2013-08-171-8/+8
|
* RF - have searchsorted copy haystack when the needle is bigger than the haystackBago Amirbekian2013-08-171-12/+14
|
* BF - cast psorter pointer to intpBago Amirbekian2013-08-171-3/+3
|
* TST - add testing of searchsorted on unaligned arraysBago Amirbekian2013-08-171-4/+61
|
* RF - allow ndarray.searchsorted to avoid making a copy for non-contiguous arraysBago Amirbekian2013-08-171-18/+26
|
* Merge pull request #3630 from rgommers/fix-medianCharles Harris2013-08-172-56/+74
|\ | | | | BUG: fix np.median so it accepts array_like input. Clean up median test...
| * BUG: fix np.median so it accepts array_like input. Clean up median tests.Ralf Gommers2013-08-172-56/+74
| |
* | Merge pull request #3628 from seberg/issue-3589Charles Harris2013-08-172-19/+48
|\ \ | |/ |/| BUG: MapIter leaked itself and arr and index objects
| * BUG: MapIter leaked itself and arr and index objectsSebastian Berg2013-08-162-19/+48
| | | | | | | | | | | | | | | | | | | | | | | | | | | | MapIterBind incorrectly behaved when failing, cleaning up most but not everything. The callers have to decref the mapiter object. Which basically means that MapIterBind does not even have to decref the new fields (they are decrefed in the mapiter dealloc anyway). Note that trying to add tests checking the refcounting failed, but the tests seem to fail at least in many cases not because of the indexing. Closes gh-3589
* | MAINT: Move NPY_BEGIN_THREADS_DEF to end of declarations.Charles Harris2013-08-161-11/+11
| | | | | | | | | | | | | | | | | | | | The is an ';' at the end of the macro definition. When used and followed by another ';', gcc thinks a statement has been made. An empty statement, but still a statement. This causes compilation failure with Python 3.4a. The ';' should probably be removed, but that was undone in 1.7 to maintain backward compatibility.
* | Change casting type for ufunc operandsJay Bourque2013-08-161-1/+1
| |
* | Bail out of loop if python error occursJay Bourque2013-08-162-6/+14
| |
* | Check for python error when needs_api flag is setJay Bourque2013-08-161-1/+6
| |
* | Update docsJay Bourque2013-08-161-3/+3
| |
* | Update docsJay Bourque2013-08-161-0/+3
| |
* | Update docsJay Bourque2013-08-161-18/+24
| |
* | Fix memory leaks and inc dtype refsJay Bourque2013-08-162-4/+10
| |
* | Add decrefs at end of function and return None objectJay Bourque2013-08-161-7/+19
| |
* | Move gil release/aquire outside of loopJay Bourque2013-08-161-10/+10
| |
* | Move NpyIter_IterNextFunc call outside of loopJay Bourque2013-08-161-7/+7
| |
* | Add test for non native bytetypesJay Bourque2013-08-161-0/+6
| |
* | Add test for empty subspaceJay Bourque2013-08-161-0/+5
| |
* | Add NpyIter_DeallocateJay Bourque2013-08-161-0/+2
| |
* | Use Py_XDECREF in fail sectionJay Bourque2013-08-161-6/+3
| |
* | Fix deprecation warningJay Bourque2013-08-161-1/+1
| |
* | Reverse commit 6ec51b3Jay Bourque2013-08-161-10/+0
| |
* | Remove unnecessary checkJay Bourque2013-08-161-5/+3
| |
* | More clean upJay Bourque2013-08-161-3/+4
| |
* | Clean up previous comment and add test for unary ufunc.Jay Bourque2013-08-162-7/+46
| |
* | Use NpyIter object to buffer each input valueJay Bourque2013-08-161-3/+99
| | | | | | | | Use NpyIter object to buffer each input value in order to handle certain situations such as operands with mixed dtypes.
* | Revert "Call iterator_loop instead of directly calling ufunc."Jay Bourque2013-08-161-55/+22
| | | | | | | | This reverts commit 2a9d5577a087e664ee047b3e099c1355000d8661.
* | Maybe this will fix the DeprecationWarningJay Bourque2013-08-161-2/+2
| |
* | Another attempt at fixing DeprecationWarningJay Bourque2013-08-161-2/+2
| |
* | Fix DeprecationWarning on python 3.3Jay Bourque2013-08-161-1/+1
| |
* | Add test for boolean indexing and boolean ufuncJay Bourque2013-08-161-0/+5
| |
* | Add test for mixed dtypesJay Bourque2013-08-161-0/+5
| |
* | Call iterator_loop instead of directly calling ufunc.Jay Bourque2013-08-161-22/+55
| | | | | | | | Call iterator_loop which will call ufunc and handle any buffering for cases with mixed dtypes.
* | Just create an array object for second operand; don't cast to dtype of first ↵Jay Bourque2013-08-161-9/+3
| | | | | | | | operand.
* | Add tests for 0-dim arrayJay Bourque2013-08-161-5/+7
| |
* | Improve comment for creating iter object from second operandJay Bourque2013-08-161-3/+7
| |
* | Let indexing code do error checking for arguments with mismatched dimensionsJay Bourque2013-08-161-14/+0
| |
* | Remove keyword support for required argumentsJay Bourque2013-08-161-5/+3
| |
* | Replace calls to New/Bind/Reset with MapIterArrayJay Bourque2013-08-162-12/+2
| |
* | Reword some comments/documentationJay Bourque2013-08-162-6/+9
| |
* | Add more commentsJay Bourque2013-08-161-19/+37
| |
* | Fix for crash when PyArray_MapIterSwapAxes is calledJay Bourque2013-08-161-0/+10
| | | | | | | | Set up default iteraxes array values in call to MapIterBind().
* | Fix build issueJay Bourque2013-08-161-1/+1
| |
* | Fix issue with 'at' method when indices are all slice objectsJay Bourque2013-08-162-2/+24
| |
* | Add more unit tests for 'at' methodJay Bourque2013-08-162-2/+6
| |
* | Code cleanup for 'at' methodJay Bourque2013-08-161-15/+11
| |
View Lorry Controller Status