On Fri, 12 Jun 2009 22:54:14 +0900, David Cournapeau <courn...@gmail.com> wrote:
On Fri, Jun 12, 2009 at 9:50 PM, Jean-Paul Calderone<exar...@divmod.com> wrote:
On Fri, 12 Jun 2009 11:33:36 GMT, Alan G Isaac <alan.is...@gmail.com> wrote:
On 6/12/2009 5:55 AM Virgil Stokes apparently wrote:
Any suggestions on installing matplotlib for Python 2.6.2 on a Windows
Vista platform?
Maintainers for some packages have run into a wall
compiling for 2.6. Matplotlib is one of these:
http://www.nabble.com/binary-installers-for-python2.6--libpng-segfault%2C-MSVCR90.DLL-and-%09mingw-td23971661.html
Another package I care about is SimpleParse, which also
found compiling for 2.6 to be impossible. I do not know
if this was the same problem or not, but it means that
SimpleParse is *still* not available as an installer for 2.6.
I assume this is of great concern to the Python community,
but I do not know where the discussion is taking place.
Some discussion has occurred in the issue tracker:
http://bugs.python.org/issue3308
http://bugs.python.org/issue6007
In general, the people responsible for how CPython builds on Windows
don't seem to consider this an issue.
We got the same problem with numpy. The good news is it is solvable
(at least partially).
http://projects.scipy.org/numpy/browser/trunk/numpy/random/mtrand/randomkit.c
Basically, there are some functions which are erroneously "declared"
in the .lib, but they don't actually exist in the MS C runtime. We
detect when the .C code is built under mingw, and add an hack to
redirect the function to the "real" function, _ftime64 (see around
line 75).
Thanks for pointing out one possible workaround. I don't think this is
always applicable, though. For example, pyOpenSSL can't be built for
Python 2.6 because OpenSSL uses localtime, another function with a similar
problem as the one with ftime. To redirect its usage, OpenSSL itself needs
to be changed and then distributed along with pyOpenSSL.
Jean-Paul
--
http://mail.python.org/mailman/listinfo/python-list
