diff options
| author | cookedm <cookedm@localhost> | 2006-07-26 11:37:51 +0000 |
|---|---|---|
| committer | cookedm <cookedm@localhost> | 2006-07-26 11:37:51 +0000 |
| commit | 598e6ff5286b0cc672b3211271ee51d7d18e0174 (patch) | |
| tree | f4fccf697b422efdb139b1fc5afd64075a19fb96 /numpy/lib/function_base.py | |
| parent | f68d43102a60376764e1226fc774548b89bcf3c8 (diff) | |
| download | numpy-598e6ff5286b0cc672b3211271ee51d7d18e0174.tar.gz | |
Fix for longdouble not working on OS X with gcc 4 (#183).
Replace the _SOFFSET_ macro (which got the address of the obval member
by casting to a PyScalarObject) with a function scalar_value, which
checks the type_num. This allows Py*ScalarObject's to have different
binary layout than PyScalarObject (PyLongDoubleObject had its obval at
a different place due to alignment issues).
Diffstat (limited to 'numpy/lib/function_base.py')
0 files changed, 0 insertions, 0 deletions