summaryrefslogtreecommitdiff
path: root/numpy
Commit message (Collapse)AuthorAgeFilesLines
* Import documentation from doc wiki (part 2, work-in-progress docstrings, but ↵Pauli Virtanen2008-10-2835-993/+3544
| | | | they are still an improvement)
* Import documentation from doc wiki (part 1)Pauli Virtanen2008-10-278-62/+82
|
* Change way maximum and minimum deal with nans.Charles Harris2008-10-213-24/+169
| | | | | | | | | | | | | | | | | | | Add ufuncs fmax and fmin. In the following, a complex number is considered a nan if the real or imaginary part is nan. This means that there are many different complex numbers that are nans and this effects the nan values returned by the maximum, minimum, fmax, and fmin ufuncs. The maximum and minimum ufuncs are the same as before unless nans are involved. In the case of nans, if both values being compared are nans, then the first is returned, otherwise the nan value is returned. This has the effect of propagating nans. The fmax and fmin ufuncs return the same values as maximum and minimum if neither value being compared is nan. In the case of nans, if both values being compared are nans, then the first is returned, otherwise the non-nan value is returned. This has the effect that nans are ignored unless both values are nan.
* Revert merge.Stefan van der Walt2008-10-204-490/+0
|
* Merge generalised ufuncs branch.Stefan van der Walt2008-10-204-0/+490
|\
| * Add missing file.Stefan van der Walt2008-08-221-0/+417
| |
| * Add tests [patch by Wenjie Fu and Hans-Andreas Engel].Stefan van der Walt2008-08-212-0/+72
| |
| * Move PyUFunc_FromFuncAndDataAndSignature to bottom of ufunc_api_order.txtStefan van der Walt2008-08-181-2/+2
| | | | | | | | [for Hans-Andreas Engel].
* | __getattribute__ : make sure than a np.void is returned when retrieving the ↵pierregm2008-10-191-12/+18
| | | | | | | | unmasked attribute of a single record.
* | * make sure we call self.failUnless instead of assertpierregm2008-10-191-5/+5
| |
* | core : * fixed make_mask_descr to recognize subdtypes pierregm2008-10-185-87/+237
| | | | | | | | | | | | | | | | | | : * fixed some corner cases of view mrecords : * fixed view : * got rid of _fieldmask for _mask : * WARNING: setting ._mask no longer calls __setmask__ : BE CAREFUL. : * prevent setting ._mask directly, as it screws up things: use .mask instead test_subclassing : * clean up
* | More unused vars tagged (lapack_lite).David Cournapeau2008-10-081-18/+18
| |
* | More unused vars tagged (fftpack).David Cournapeau2008-10-081-6/+6
| |
* | More unused vars tagged.David Cournapeau2008-10-081-15/+16
| |
* | Tag unused variables in numpy.lib C code.David Cournapeau2008-10-081-9/+9
| |
* | More unused warning handled.David Cournapeau2008-10-083-16/+19
| |
* | Fix missing initializers warnings.David Cournapeau2008-10-087-109/+537
| |
* | Use NPY_UNUSED in numpy.core sources.David Cournapeau2008-10-0811-170/+204
| |
* | Add utils.h, with macro to tag a variable as unused.David Cournapeau2008-10-081-0/+19
| |
* | Fix function prototypes decl in check_func to avoid warning with ↵David Cournapeau2008-10-051-2/+2
| | | | | | | | -Wstrict-prototypes.
* | Merged revisions 5882-5911 via svnmerge from David Cournapeau2008-10-058-905/+891
|\ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | http://svn.scipy.org/svn/numpy/trunk ........ r5886 | charris | 2008-10-02 03:05:29 +0900 (Thu, 02 Oct 2008) | 4 lines Make some error messages more informative. Improve error handling. Make continuation lines work. ........ r5887 | charris | 2008-10-02 03:06:04 +0900 (Thu, 02 Oct 2008) | 2 lines Small cleanup to clarify repeated string. ........ r5888 | charris | 2008-10-02 03:08:41 +0900 (Thu, 02 Oct 2008) | 6 lines Cleanup ufunc loops. At this point loops are separated into variable kinds, so there is a fair amount of duplication. I will probably merge loops that look the same in a later commit. There are no changes to current behavior of loops, this will also be changed in later work to deal with nans and such. ........ r5889 | oliphant | 2008-10-03 05:27:17 +0900 (Fri, 03 Oct 2008) | 1 line Fix problem with subclasses of object arrays. ........ r5896 | cdavid | 2008-10-03 15:50:32 +0900 (Fri, 03 Oct 2008) | 1 line Update the minimum version for numscons: had to change to cope with Chuck changes to conv_template.py. ........ r5897 | cdavid | 2008-10-03 15:51:03 +0900 (Fri, 03 Oct 2008) | 1 line Update doall script: take the python version to build binaries from the command line instead of global variable. ........ r5906 | oliphant | 2008-10-04 00:55:52 +0900 (Sat, 04 Oct 2008) | 1 line Fix ticket #925 ........
| * | Fix ticket #925Travis Oliphant2008-10-034-37/+116
| | |
| * | Update the minimum version for numscons: had to change to cope with Chuck ↵David Cournapeau2008-10-031-1/+1
| | | | | | | | | | | | changes to conv_template.py.
| * | Fix problem with subclasses of object arrays.Travis Oliphant2008-10-022-2/+15
| | |
| * | Cleanup ufunc loops.Charles Harris2008-10-011-890/+790
| | | | | | | | | | | | | | | | | | | | | At this point loops are separated into variable kinds, so there is a fair amount of duplication. I will probably merge loops that look the same in a later commit. There are no changes to current behavior of loops, this will also be changed in later work to deal with nans and such.
| * | Small cleanup to clarify repeated string.Charles Harris2008-10-011-1/+1
| | |
| * | Make some error messages more informative.Charles Harris2008-10-011-13/+28
| | | | | | | | | | | | | | | Improve error handling. Make continuation lines work.
* | | Generate math_c99.inc in numscons build.David Cournapeau2008-10-031-0/+1
| | |
* | | Remove old configuration checks, supersded by new math config.David Cournapeau2008-10-031-32/+0
| | |
* | | Add check for C99 macros related IEEE-754.David Cournapeau2008-10-031-0/+12
| | |
* | | Check for float/long double C99 functions.David Cournapeau2008-10-031-1/+14
| | |
* | | Check for some optional, C99 double math functions.David Cournapeau2008-10-031-0/+7
| | |
* | | Fix typo in check_funcs.David Cournapeau2008-10-031-1/+1
| | |
* | | Add an help function to check a list of functions in scons build.David Cournapeau2008-10-031-0/+9
| | |
* | | Start updating numscons configuration for new math config.David Cournapeau2008-10-031-2/+10
| | |
* | | Merged revisions 5861-5881 via svnmerge from David Cournapeau2008-09-308-185/+223
|\ \ \ | |/ / | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | http://svn.scipy.org/svn/numpy/trunk ........ r5862 | ptvirtan | 2008-09-24 20:50:38 +0900 (Wed, 24 Sep 2008) | 1 line Fix python2.5 dependency in lookfor ........ r5869 | pierregm | 2008-09-29 02:27:37 +0900 (Mon, 29 Sep 2008) | 3 lines core: * added __rmul and __radd__ * fixed concatenate for flexible-dtype ........ r5878 | ptvirtan | 2008-09-30 02:23:43 +0900 (Tue, 30 Sep 2008) | 1 line Umath tests: remove signed zero check on branch cut for log* as it fails on some platforms. ........ r5879 | pierregm | 2008-09-30 05:24:29 +0900 (Tue, 30 Sep 2008) | 1 line use if ...: raise AssertionError instead of assert ........ r5880 | pierregm | 2008-09-30 05:24:56 +0900 (Tue, 30 Sep 2008) | 1 line replaced assert with self.failUnless ........
| * | replaced assert with self.failUnlesspierregm2008-09-294-166/+166
| | |
| * | use if ...: raise AssertionError instead of assertpierregm2008-09-291-7/+14
| | |
| * | Umath tests: remove signed zero check on branch cut for log* as it fails on ↵Pauli Virtanen2008-09-291-3/+8
| | | | | | | | | | | | some platforms.
| * | core:pierregm2008-09-282-1/+22
| | | | | | | | | | | | | | | * added __rmul and __radd__ * fixed concatenate for flexible-dtype
| * | Fix python2.5 dependency in lookforPauli Virtanen2008-09-241-8/+13
| | |
* | | Enhance code_generator to allow continuation lines.Charles Harris2008-09-282-14/+29
| | | | | | | | | | | | | | | Small clarification in arraytypes.inc.src. Practice merging ;)
* | | Merged revisions 5844-5860 via svnmerge from David Cournapeau2008-09-232-1/+16
|\ \ \ | |/ / | | | | | | | | | | | | | | | | | | | | | | | | http://svn.scipy.org/svn/numpy/trunk ........ r5860 | stefan | 2008-09-23 05:28:00 +0900 (Tue, 23 Sep 2008) | 2 lines Ignore unused converters in `loadtxt`. ........
| * | Ignore unused converters in `loadtxt`.Stefan van der Walt2008-09-222-1/+16
| | |
* | | Do not use MSVC workaround anymore, since we now can test for function even ↵David Cournapeau2008-09-211-12/+4
| | | | | | | | | | | | when they are intrinsincs.
* | | define -> pragma, stupid mistake.David Cournapeau2008-09-211-2/+2
| | |
* | | Handle msvc intrisincs in check_funcs_once.David Cournapeau2008-09-211-0/+6
| | |
* | | Handle msvc intrisincs in check_func.David Cournapeau2008-09-211-0/+7
| | |
* | | More intrinsics.David Cournapeau2008-09-201-1/+1
| | |
* | | Fix for isfinite on VS 2003.David Cournapeau2008-09-201-1/+3
| | | | | | | | | | | | | | | | | | VS 2003 seems to think it is ok to simplify x-x to 0 for float, but this is wrong for NaN and Inf. To alleviate, we force the operation to occur with (x) + (-x).
View Lorry Controller Status