summaryrefslogtreecommitdiff
path: root/numpy
Commit message (Collapse)AuthorAgeFilesLines
...
* BUG: core: ensure keywords[] list is NULL terminated in array_choose (might ↵Pauli Virtanen2010-05-051-1/+1
| | | | be related to #1476)
* BUG: use PyOS_string_to_double on Python 2.7Pauli Virtanen2010-05-051-2/+2
| | | | | | | | | | Related to #1345. This works around PyOS_ascii_strtod raising a PendingDeprecationWarning when GIL is released, which results to a crash. The full fix would be to re-acquire GIL, but I believe we will not want to do that for performance reasons, as it doesn't otherwise seem to be necessary.
* ENH: Do some housekeeping for the newly added slogdet function.Charles Harris2010-05-051-0/+2
|
* ENH: Add slogdet to the linalg module. The patch is from njs withCharles Harris2010-05-053-11/+97
| | | | slogdet substituted for sign_log_det. Closes ticket #1402.
* ENH: Add test of polyder return type.Charles Harris2010-05-051-0/+8
|
* BUG: Make polyder return a poly1d for the zeroeth order derivative whenCharles Harris2010-05-051-8/+9
| | | | the input is a poly1d. Fixes ticket #1249.
* BUG: Fix datetime_data for python versions >= 2.7.Charles Harris2010-05-041-1/+4
|
* ENH: Simplify list destructor.Charles Harris2010-05-041-6/+6
|
* BUG: The builtin file function goes away in python 3k, use open instead.Charles Harris2010-05-031-1/+1
|
* BUG: Fix use of int type where Py_ssize_t is appropriate.Charles Harris2010-05-031-1/+1
|
* ENH,BUG: Simplify array_choose with NpyArg_ParseKeywords. Fix some usesCharles Harris2010-05-031-17/+10
| | | | of int where Py_ssize_t should be used.
* ENH, BUG: PyCObject will be deprecated in python 2.7. So use the NpyCapsuleCharles Harris2010-05-0323-241/+71
| | | | | | | compatibility functions in npy_3kcompat.h to replace the current calls. This gets rid of a number of version checks and is easier to maintain. Fix bug that was present in the ufunc _loop1d_list_free destructor in the python3k case.
* BUG: loadtxt should handle universal newlines.Stefan van der Walt2010-05-022-1/+11
|
* ENH: core: add .dot() method to ndarrays; a.dot(b) == np.dot(a, b)Pauli Virtanen2010-04-303-2/+36
|
* BUG: In datetime, avoid overflow by casting multipliers to npy_int64.Stefan van der Walt2010-04-273-52/+77
|
* BUG: Fix missing import for datatime_data.Stefan van der Walt2010-04-272-0/+8
|
* Fixed .var for arrays with 1 more valid value than ddofspierregm2010-04-272-2/+15
|
* Add fixme note.Charles Harris2010-04-271-0/+1
|
* BUG: core: fix a reference count bug in umath:InitOtherOperators (cf #1464)Pauli Virtanen2010-04-261-1/+0
| | | | Thanks to mras
* ENH: Simplify some functions using NpyArg_ParseKeywords.Charles Harris2010-04-261-18/+6
|
* ENH: Add NpyArg_ParseKeywords helper function.Charles Harris2010-04-261-0/+27
|
* BUG: fix reference count bug in set_numeric_ops (fixes #1464, #1462)Pauli Virtanen2010-04-251-0/+1
|
* ENH: Add some tests for ndarray.resize.Charles Harris2010-04-251-0/+19
|
* BUG: Catch resize shape conversion errors.Charles Harris2010-04-251-5/+8
|
* BUG: Fix the fix so resize works like it is supposed to.Charles Harris2010-04-251-6/+25
|
* ENH: Remove order keyword from ndarray.resize. This is pretty muchCharles Harris2010-04-255-75/+41
| | | | Scott Sinclair's patch from ticket #840.
* ENH: Remove some unused variable warnings.Charles Harris2010-04-253-11/+7
|
* BUG: Fix missing "%s" in error message format. Remove unused variable.Charles Harris2010-04-251-2/+2
|
* BUG: Fix wrong destructor type in NpyCapsule_FromVoidPtrAndDesc.Charles Harris2010-04-251-1/+2
|
* FIX: Workaround python3.1 callable not working with PyCapsule.Charles Harris2010-04-251-1/+4
|
* ENH: Apply old patch hanging around in my directory. I don't recall where itCharles Harris2010-04-251-0/+6
| | | | came from, but it seems appropriate.
* correct a bug in fix() that was introduced in r8293Darren Dale2010-04-182-1/+2
|
* BUG: Use deprecated decorator in testing ufunclike.log2.Charles Harris2010-04-142-11/+9
|
* BUG: Python 2.4 doesn't support "with" statement, use try instead.Charles Harris2010-04-131-1/+3
|
* BUG: Deprecate ufunclike.log2 and take it off the __all__ list. It was shadowingCharles Harris2010-04-131-1/+6
| | | | the ufunc of the same name.
* ENH: Add test of sign ufunc removed from test_ufunclike.Charles Harris2010-04-131-0/+48
|
* ENH: Rewrite doctest in test_ufunclike.py as normal nose tests. Remove test ofCharles Harris2010-04-131-74/+60
| | | | sign ufunc, it belongs elsewhere.
* ENH: Add more tests for new memmap object attributes.Charles Harris2010-04-131-2/+16
|
* BUG: Need to use os.path.abspath on file handle name also.Charles Harris2010-04-121-1/+1
|
* ENH: Make the new memmap filename attribute store the full path to the file.Charles Harris2010-04-121-1/+14
| | | | Add offset and mode attributes to the memmap class.
* ENH: Add filename attribute to memmap.Charles Harris2010-04-122-0/+14
|
* ENH: Add test for log1p on intel 32 bits for certain small values of p.Charles Harris2010-04-051-0/+4
|
* ENH: Try to make log1p a bit more resistant to compiler shenanigans.Charles Harris2010-04-051-6/+9
| | | | STY: Couple of style fixups.
* ENH: Use standard npy constant definitions in log2, exp2 functions.Charles Harris2010-04-051-17/+14
| | | | | STY: Some cleanups, fix goto into if statement. BUG: Use proper name for expm1, not exp1m.
* ENH: Make npy_log2_1p and npy_exp2_1m use standard functions.Charles Harris2010-04-051-59/+49
| | | | | This avoids some problems with gcc optimizations on 32 intel systems. In any case, the problem is pushed back to location single functions.
* ENH: core: add sanity checks and warnings to PEP3118 -> ndarray conversionPauli Virtanen2010-04-041-0/+14
|
* ENH: core: improve the way trailing padding is dealed with in PEP 3118 ↵Pauli Virtanen2010-04-042-16/+92
| | | | format strings
* ENH: core: improve parsing of byte order from PEP 3118 format stringsPauli Virtanen2010-04-042-7/+20
|
* ENH: core: improve PEP 3118 parser's alignment handlingPauli Virtanen2010-04-042-12/+72
| | | | Now, padding will be more forcefully inserted for native-aligned items.
* BUG: core: fix bugs in PEP 3118 format string parsingPauli Virtanen2010-04-042-6/+11
| | | | | - Handle consecutive characters (eg. 'xxxx') properly - Compute native padding correctly