summaryrefslogtreecommitdiff
path: root/numpy
Commit message (Collapse)AuthorAgeFilesLines
* ENH: add fft/random support.David Cournapeau2011-08-295-3/+73
|
* FEAT: add numpy.linalg.David Cournapeau2011-08-292-0/+46
|
* FEAT: numpy.core and numpy.lib both buildDavid Cournapeau2011-08-294-2/+243
|
* BENTO: add complex checks.David Cournapeau2011-08-292-0/+48
|
* BENTO: check for inline support.David Cournapeau2011-08-292-0/+39
|
* BENTO: add math runtime checks.David Cournapeau2011-08-292-0/+238
|
* BENTO: add signal/smp stuff in configure.David Cournapeau2011-08-291-0/+46
|
* BENTO: add basic type checks.David Cournapeau2011-08-292-0/+163
|
* BENTO: check for C99 formats + add check_declaration.David Cournapeau2011-08-292-0/+35
|
* BENTO: start support for numpyconfig.David Cournapeau2011-08-291-1/+26
|
* BENTO: start numpy.core configuration + waf_utils helpers.David Cournapeau2011-08-291-0/+0
|
* BENTO: start numpy.core configurationDavid Cournapeau2011-08-293-0/+136
| | | | We also add configuration helpers in numpy.build_utils.
* ENH: missingdata: Make PyArray_Converter and PyArray_OutputConverter safer ↵Mark Wiebe2011-08-276-24/+96
| | | | | | | | for legacy code These functions now reject inputs with NA, and there are alternative functions PyArray_AllowNAConverter and PyArray_OutputAllowNAConverter that functions should use when they intend to support NA.
* ENH: nditer: Change the Python nditer exposure to automatically add ↵Mark Wiebe2011-08-272-5/+50
| | | | NPY_ITER_USE_MASKNA
* ENH: missingdata: Make comparisons with NA return NA(dtype='bool')Mark Wiebe2011-08-273-63/+85
|
* BLD: core: onefile build fix and Python3 compatibility changeMark Wiebe2011-08-272-3/+4
|
* TST: dtype: Adjust void dtype test to pass without raising a zero-size exceptionMark Wiebe2011-08-271-1/+1
|
* TST: missingdata: Write some tests for the np.any and np.all NA behaviorMark Wiebe2011-08-272-7/+87
|
* ENH: missingdata: Make numpy.all follow the NA && False == False ruleMark Wiebe2011-08-279-95/+172
|
* ENH: missingdata: Make numpy.all follow the NA || True == True ruleMark Wiebe2011-08-2710-45/+508
|
* DOC: missingdata: Add introductory documentation for NA-masked arraysMark Wiebe2011-08-271-0/+38
|
* ENH: core: Rename PyArrayObject_fieldaccess to PyArrayObject_fieldsMark Wiebe2011-08-2719-135/+135
| | | | Because it's shorter and seems like a decent name.
* DOC: missingdata: Add example of a C-API function supporting NA masksMark Wiebe2011-08-271-1/+2
|
* ENH: missingdata: Finish adding C-API access to the NpyNA objectMark Wiebe2011-08-276-47/+66
|
* ENH: missingdata: Expose Npy_NA as a global singleton, same as Py_NoneMark Wiebe2011-08-274-5/+6
|
* ENH: missingdata: Rename na_singleton.[ch] to na_object.[ch]Mark Wiebe2011-08-2715-15/+16
|
* BUG: missingdata: Add support for NA masks to the order='K' case of np.ravelMark Wiebe2011-08-272-33/+57
|
* ENH: missingdata: Add maskna= and ownmaskna= parameters to np.asarray and ↵Mark Wiebe2011-08-276-38/+182
| | | | | | | | friends Also fix some array() NA mask construction issues and make sure the base object doesn't collapse past the owner of the NA mask being viewed in addition to the data.
* ENH: missingdata: Add maskna= flag to np.identity constructorMark Wiebe2011-08-272-4/+13
|
* ENH: missingdata: Add maskna= flag to np.eye constructorMark Wiebe2011-08-273-11/+24
|
* BUG: umath: Make the ufunc follow the actual priority for __r<op>__Mark Wiebe2011-08-271-18/+22
| | | | | Previously it returned not implemented just based on the existence of the array priority, it didn't actually check the priority value.
* ENH: core: Add static caching of the callables for C to core._method callsMark Wiebe2011-08-272-19/+147
|
* ENH: missingdata: Add skipna=, keepdims= parameters to methodsMark Wiebe2011-08-276-225/+146
| | | | | | Also fix some memory leaks, improve some type resolution code. The methods still have some issues with array subtypes that needs working through.
* ENH: missingdata: Add support for identity-less skipna reductions with ndim > 1Mark Wiebe2011-08-273-10/+310
|
* ENH: missingdata: Future-proof AssignNA and AssignMaskNA for later multi-NA ↵Mark Wiebe2011-08-277-30/+39
| | | | support
* ENH: missingdata: Move ReduceMaskNAArray out of the public APIMark Wiebe2011-08-274-17/+59
| | | | | | Keep this function private, to keep flexibility open for longer. Also added what will likely be the necessary parameters for future where mask and multi-NA expansion.
* ENH: missingdata: Add wheremask to PyArray_ContainsNAMark Wiebe2011-08-2710-52/+243
| | | | | | Use this to make masked assignment just check the elements its copying for NA, so that the source array can have NAs, just not where the mask says.
* ENH: ufunc: Remove CreateReduceResult and InitializeReduceResult from the APIMark Wiebe2011-08-273-139/+90
| | | | | | Keep them internal longer, to let the details bake some more. PyArray_ReduceWrapper is in the API so both multiarray and umath can use it.
* BUG: dtype: Mitigate crash bug for some kinds of dtype inputsMark Wiebe2011-08-273-5/+21
| | | | | | | This doesn't fix it in general, but does for the specific case. The particular code in question is calling from C into Python to do parsing. I think this is a bad idea, and should be converted to straightforward C code instead.
* ENH: missingdata: Make ndarray.view validate the maskna= parameter betterMark Wiebe2011-08-272-4/+38
|
* ENH: missingdata: Add maskna= parameter to np.copy and ndarray.copyMark Wiebe2011-08-278-156/+319
|
* ENH: ufunc: Add a mask dtype parameter to the masked ufunc loop selectorMark Wiebe2011-08-274-1/+15
| | | | This is to allow for future expansion to multi-NA and struct-NA.
* ENH: umath: Switch PyUFunc_Reduce to call PyArray_ReduceWrapper to simplify codeMark Wiebe2011-08-2710-621/+480
|
* ENH: umath: Add checking for reorderable ufuncs, add PyArray_ReduceWrapper ↵Mark Wiebe2011-08-279-53/+261
| | | | | | | | to API np.divide, for instance, isn't commutative, so the computation should not change the order of computation. The reduce function disables multiple-axis reductions on ufuncs which are not reorderable.
* ENH: ufunc: Separate type resolution from loop selectionMark Wiebe2011-08-276-745/+757
| | | | | | | | | | | | | This change sets the stage for being able to produce much higher performance loops by specializing based on any fixed strides. To do this, type resolution must happen first to determine the dtypes, then the iterator needs to be created, then finally the loop needs to be selected for the given dtypes and the strides that will be fixed throughout the iteration. The way type resolution and loop selection where tied together before precluded this possibility, but with the backwards-compatible design change in this commit, such specializations will be possible in the future.
* BUG: missingdata: Fix long double printing of NAsMark Wiebe2011-08-271-3/+3
|
* BUG: missingdata: np.isna function wasn't accepting object arraysMark Wiebe2011-08-271-1/+7
|
* ENH: missingdata: Improve error message when assigning NA to non-NA-masked ↵Mark Wiebe2011-08-273-4/+73
| | | | arrays
* BUG: repr: Make NA line up in the float array repr like inf and nanMark Wiebe2011-08-272-3/+3
|
* ENH: missingdata: Change NA repr to use strings like the dtype repr doesMark Wiebe2011-08-275-53/+77
|
View Lorry Controller Status