Permalink
Commits on Oct 29, 2011
Commits on Oct 23, 2011
  1. BUG: Fix maximum/minimum for object arrays in Python 3.

    charris committed Oct 23, 2011
    Use PyObject_RichCompareBool instead of PyObject_Cmp.
  2. STY: Some coding style fixups.

    charris committed Oct 23, 2011
  3. ENH: Add 'pgfortran' as a valid compiler in the Portland Group

    tim cera committed with charris Oct 20, 2011
    suite of fortran compilers. Adjust the version string command.
Commits on Oct 17, 2011
Commits on Oct 10, 2011
Commits on Oct 8, 2011
Commits on Oct 4, 2011
  1. ENH: Incorporated review advice and changed macro hack to MinGW hint.

    87 committed with charris Sep 17, 2011
    When doing the typedef change, it conflicted with existing typedefs.
    I realized that MinGW did not know which MSVC runtime it would link with,
    but it had conditional definitions for MSVC 8 and up, which can be activated
    by defining __MSVCRT_VERSION__. So I added it to the compiler macros, which
    made the fallback hack and extra typedef unnecessary.
  2. BUG: Also build MSVC debug library if MINGW_USE_CUSTOM_MSVCR is defin…

    87 committed with charris Sep 14, 2011
    …ed (best effort).
  3. ENH: Clean up symbol definition file.

    87 committed with charris Sep 14, 2011
Commits on Oct 2, 2011
  1. BUG: Fixed reference count bug where scalar value assignment by field…

    cfarrow committed with charris Aug 6, 2011
    … to an array would increment the reference count of the scalar.
  2. BUG: fixed reference counting bug in PyArray_FromScalar that occurs w…

    cfarrow committed with charris Aug 27, 2011
    …hen using an extended dtype
Commits on Oct 1, 2011
  1. PyArray_FromBuffer: Allow creating arrays from empty buffers or empty…

    Hrvoje Niksic committed with charris Sep 19, 2011
    … slices.
  2. Merge branch 'pull-161'

    charris committed Oct 1, 2011
    * pull-161:
      BUG: Attempt Y2038 fix, and restrict local timezone parsing to 1970 and later, because of Win32
      BUG: Attempt to fix the Y2038 problem parsing dates >= 2038
      TST: Add tests for the Y2038 problem
      BUG: Remove floor_divide support for timedelta64 // timedelta64
      DOC: Update year range in LICENSE.txt copyright
      TST: Add true_divide and floor_divide tests for the mm_d timedelta case
      STY: Fix a few cases of spaces not following conventions
      BUG: Add %lld exception for Python 3.1 as well
      WRN: Remove many warnings when building at a higher warning level
      BUG: datetime: Fix timedelta / timedelta -> float64 on Python 3
      BUG: datetime: Fix str() function of datetime arrays
      BUG: datetime: Make datetime_as_string produce a UNICODE array instead of STRING array on Python3
  3. BUG: Attempt Y2038 fix, and restrict local timezone parsing to 1970 a…

    mwiebe committed with charris Sep 30, 2011
    …nd later, because of Win32
    
    On MS Windows, dates earlier than 1970 cause mktime to error. This patch
    also applies to parsing the same Y2038 hack that was done for printing.
  4. BUG: Attempt to fix the Y2038 problem parsing dates >= 2038

    mwiebe committed with charris Sep 30, 2011
    Also incorporated part of Han Genuit's pull request #23, to try
    and reduce the difficulty dealing with the merge conflict that
    will result from these concurrent changes