- 480b20b Update revision to beta 1. by David Cournapeau ยท 17 years ago v1.3.0b1
- 4d3cf3d Remove long double test for repr round-trip. by David Cournapeau ยท 17 years ago
- e672f36 Fix exception message when the shape of a subarray is not a tuple. by David Cournapeau ยท 17 years ago
- c931ce6 Use Py_ssize_t for PyDict_Next. by David Cournapeau ยท 17 years ago
- 12038db Remove obsolete comments. by David Cournapeau ยท 17 years ago
- 6e9c1b7 Fix English. by David Cournapeau ยท 17 years ago
- 4f35a21 Add one nested compound dtype. by David Cournapeau ยท 17 years ago
- 04bad09 BUG: hashing dtype failed for subarray which shape was an integer instead of a tuple. by David Cournapeau ยท 17 years ago
- baa4d11 Fix missing declaration for status code. by David Cournapeau ยท 17 years ago
- 3963202 Use python exception system instead of ad-hoc error code. by David Cournapeau ยท 17 years ago
- f9e7ba9 Fix comment. by David Cournapeau ยท 17 years ago
- a5ef4a3 Add some documentation and make function names consistent. by David Cournapeau ยท 17 years ago
- c08ee20 Implement subarray walking for dtype hashing. by David Cournapeau ยท 17 years ago
- 8d3c1df Walking into fields to compute hash of compound types. by David Cournapeau ยท 17 years ago
- 617dc9d Start implementing 'real' hashing protocol: works for builtin. by David Cournapeau ยท 17 years ago
- 10cccca Using separate compilation unit does not work for array API. by David Cournapeau ยท 17 years ago
- 1bebeca Start implementing hashing protocol for type descr by David Cournapeau ยท 17 years ago
- ebc31e8 Add some unit tests for the hashing protocol of dtype (fail currently). by David Cournapeau ยท 17 years ago
- 576a678 Make logseries test deterministic. by Charles Harris ยท 17 years ago
- a1e2f9b Fix tickets #921 and #923. Add regression tests. by Charles Harris ยท 17 years ago
- 016afc5 Give assert_ a default message of ''. by Charles Harris ยท 17 years ago
- 5f98923 Add macros for ia64 by Charles Harris ยท 17 years ago
- 74b16f6 Revert "Fix wrong formatting: Py_ssize_t vs npy_intp." by David Cournapeau ยท 17 years ago
- ff7d512 Check for the inline keywork in numscons build. by David Cournapeau ยท 17 years ago
- 2626cea Add inline keyword check. by David Cournapeau ยท 17 years ago
- 1d5343b Use setup_common for scons build. by David Cournapeau ยท 17 years ago
- 6db55e7 Put func name -> conf string convertion in setup_common. by David Cournapeau ยท 17 years ago
- ce0dd44 Put math funcs to test in separate file to share with numscons build. by David Cournapeau ยท 17 years ago
- ebd770a Back out VS 2003 hack - it is too fragile. by David Cournapeau ยท 17 years ago
- 016763c Fix wrong formatting: Py_ssize_t vs npy_intp. by David Cournapeau ยท 17 years ago
- 3042a38 Skip graciously when ctypes is not available. by David Cournapeau ยท 17 years ago
- 6f60d2c Make the gfortran/vs2003 hack source file known to distutils. by David Cournapeau ยท 17 years ago
- 6485a8e numpy.ma doc update by pierregm ยท 17 years ago
- a082b6c Ugly hack to workaround some bug in mingw to make VS 2003 work with gfortran. by David Cournapeau ยท 17 years ago
- a2c6798 Add a function to get MSVC version. by David Cournapeau ยท 17 years ago
- 43399b7 Remove pzero/pinf init functions, and use NPY_* macros instead. by David Cournapeau ยท 17 years ago
- 8f15bfc Use NPY_NAN instead of computing a nan - also fix #920, since NPY_NAN is a positive nan. by David Cournapeau ยท 17 years ago
- 5d2998a Add a regression test to check our nan is a positive nan. by David Cournapeau ยท 17 years ago
- 4e0a672 Do not use NAN (GNU specific extension) but our portable NPY_NAN instead. by David Cournapeau ยท 17 years ago
- 987d5e0 Do not pull the whole ndarrayobject header in npy math, to avoid unused _import_array warning, and because we don't need it anyway. by David Cournapeau ยท 17 years ago
- 07e396f Define NPY_INLINE as a (public) macro for portable inline, and use it in npy_math.h, to avoid spurious warnings about unused static functions. by David Cournapeau ยท 17 years ago
- dc8294d Move bit-width types to npy_common, so that we can use it in pure C code (code which does not use the python API), npymath in particular, without pulling the whole ndarray headers. by David Cournapeau ยท 17 years ago
- f87fb9c Define inline to something we can use internally. by David Cournapeau ยท 17 years ago
- 354e0eb Add check for (C) inline. by David Cournapeau ยท 17 years ago
- 21cbf24 Change complex version of sign to return nan for nan input. Fix test to by Charles Harris ยท 17 years ago
- 5847959 fix remaining tests to reflect new behavior of sign(nan) by Charles Harris ยท 17 years ago
- 670a4e7 Fix a test failure due to changes in the sign ufunc. by Charles Harris ยท 17 years ago
- a4816bc Change behavior of sign ufunc so it returns nan for nan instead of zero. This by Charles Harris ยท 17 years ago
- 1be8c35 test_umath: remove unnecessary test from #1008 test set by Pauli Virtanen ยท 17 years ago
- 7382496 Fix ticket #1048. by Charles Harris ยท 17 years ago
- d3a8a63 Set formatting references for nan/inf complexes depending on the python version, since python 2.6 have revamps its own formatting support. by David Cournapeau ยท 17 years ago
- bb41c56 Fix wrongly inserted * character for the real part of complex number with nan/inf. by David Cournapeau ยท 17 years ago
- 92e09e4 Fix C99 declaration style. by David Cournapeau ยท 17 years ago
- 82803fe Fix double comma syntax error. by David Cournapeau ยท 17 years ago
- 6433987 Remove NumpyOS_init, obsoleted by our NPY_NAN and co macros. by David Cournapeau ยท 17 years ago
- 3258937 Fix wrong cast for zero macros. by David Cournapeau ยท 17 years ago
- 7c1192a Add macros for positive and negative inf. by David Cournapeau ยท 17 years ago
- 8921d5d Use functions instead of anonymous union that some compilers don't like. by David Cournapeau ยท 17 years ago
- 5333436 Add portable equivalent of NAN/INFINITY. by David Cournapeau ยท 17 years ago
- 338d5c5 Revert adding '*' character for nan and inf before the imaginary tag for python >= 2.6. by David Cournapeau ยท 17 years ago
- 97c8aa9 numscons build for umath_tests. by David Cournapeau ยท 17 years ago
- 44bd513 Fixing #1045: don't output * in between of inf/nan and j by Pauli Virtanen ยท 17 years ago
- a49c262 Fix nan/inf handling when formatting complex numbers. by David Cournapeau ยท 17 years ago
- 4a45e03 Fixed #950: fix bug in dotblas 0D input detection. by Pauli Virtanen ยท 17 years ago
- d771e4e Fixed #836: ensure z**3 = z*z*z, z**2 == z*z and z**1 == z for complex pow+inf. by Pauli Virtanen ยท 17 years ago
- 9d6b9b9 Fixed #745: make assert_array_almost_equal work with object arrays by Pauli Virtanen ยท 17 years ago
- de62644 Apply patch from #953 to fix two doctests. by David Cournapeau ยท 17 years ago
- b51762e Fix docstring for amax/amin. by David Cournapeau ยท 17 years ago
- 7d75121 Fix and re-enable regression test #947. by David Cournapeau ยท 17 years ago
- d9b8b76 Fix #947. by David Cournapeau ยท 17 years ago
- e62aa7e Use a failure label when _array_fromobject fails. by David Cournapeau ยท 17 years ago
- 1ace444 Add regression tesst for #947. by David Cournapeau ยท 17 years ago
- 46b375c Applied Anand's patch fixing Ticket #986 regarding the domain of the Von Mises random variable generator. Regression test added. by dhuard ยท 17 years ago
- dd8299b BUG: when a new type is required, and the fast clip function is updated, make sure the function pointer is not NULL - if it is, fall back to slow array. by David Cournapeau ยท 17 years ago
- 0588cb8 Fix import issue. by David Cournapeau ยท 17 years ago
- 1d4365c Re-enable long double repr tests, as long double formatting has been fixed on windows. by David Cournapeau ยท 17 years ago
- 4b6277c Sort index of find_duplicated before testing, because the exact order is undefined for identical entries. by David Cournapeau ยท 17 years ago
- c2bb0f9 Remove dead code. by David Cournapeau ยท 17 years ago
- 372e31e Re-enable generalized ufunc tests on win32, since the corresponding module now builds correctly. by David Cournapeau ยท 17 years ago
- d4722fa Do not hardcode string for savetxt testing, as the exact representation depends on the platform. by David Cournapeau ยท 17 years ago
- 03c47e4 BUG: Rewrite test_gzip_loadtxt to avoid NamedTemporaryFile which is unusable with windows IO semantics + add another test for compressed string argument to loadtxt. by David Cournapeau ยท 17 years ago
- 2ef6634 Close file handle in loadtxt if we opened the file ourselves. by David Cournapeau ยท 17 years ago
- 969ef15 Make the type cast clearer + comment. by David Cournapeau ยท 17 years ago
- 3b2ea6c More signed/unsigned comp fixes. by David Cournapeau ยท 17 years ago
- 6a2a979 More unsigned comparison warning fixes. by David Cournapeau ยท 17 years ago
- e16daf2 More signed/unsigned fixes. by David Cournapeau ยท 17 years ago
- 54608a8 Fix signed/unsigned warnings by doing explicit cast (as done by the compiler implicitely anyway). by David Cournapeau ยท 17 years ago
- eefc141 Fix polyint to work correctly with float, complex, and int inputs. by Charles Harris ยท 17 years ago
- 4a63253 Fix ticket #944. by Charles Harris ยท 17 years ago
- 3926a6c Add assert_ as a release safe version of assert for running tests. by Charles Harris ยท 17 years ago
- 5a9fe50 Add test for array priorities involving scalars. by Charles Harris ยท 17 years ago
- 9673a83 Fix type in test_wrap_with_iterable. by Charles Harris ยท 17 years ago
- a947249 Cleanup the _find_array_wrap code a bit. by Charles Harris ยท 17 years ago
- cb9f562 Add test for wrap with iterable. Thanks to Darren Dale for the patch. by Charles Harris ยท 17 years ago
- 677173f Apply patch in ticket #826 to fix problem with subtypes. by Charles Harris ยท 17 years ago
- e1c91c7 Apply patch from Darren Dale to fix __array_wrap__ problem. by Charles Harris ยท 17 years ago
- 919f279 Apply patch from ticket #1040 fixing some 64 bit issues. Thanks to jmiller. by Charles Harris ยท 17 years ago
- 41be2e3 Make printout of precision test fail more informative. by Charles Harris ยท 17 years ago
- 090ce2a Merge reviewed docstrings from editor. by Stefan van der Walt ยท 17 years ago
- 06ffe4d Use NPY_PI instead of C math lib one, which is not always available. by David Cournapeau ยท 17 years ago
