[SciPy-user] Testing build before installing (numpy and scipy)

Robert Kern robert.kern@gmail....
Wed Apr 16 17:02:16 CDT 2008

On Wed, Apr 16, 2008 at 3:35 PM, Warren Weckesser
<warren.weckesser@gmail.com> wrote:
> I am have a Mac running OSX 10.4.  My current installation of numpy and
> scipy is broken, so I want to clean up and start from scratch. I just
> installed Python 2.5.2 using a binary from the python web site.  I have
> several questions:
> 1.  numpy's setup.py doesn't have an "uninstall" command.  Is that normal?


> What is the standard way to remove something that I installed using
> setup.py?  I want to clean out my Python2.4 broken installations.

Delete the numpy/ directory in site-packages/ and delete the f2py
script wherever it got installed to. It is probably in

> 2.  Having just installed python 2.5.2, I ran "setup.py build" in the
> numpy-1.0.4 directory.  It generated lots of output, but I don't know if
> everything built correctly.  There are some error messages about
> _configtest.c having errors.

Don't worry about these. In order to figure out if your system
supports certain features, we try to compile and execute a number of
small C programs. If the compilation fails, then your system doesn't
support that feature; that's fine, we just make the appropriate
configuration setting.

> 3. The instructions at http://www.scipy.org/Installing_SciPy/Mac_OS_X
> suggest that the command  "export MACOSX_DEPLOYMENT_TARGET=10.4" be given
> before building scipy for OSX 10.4.  It is not clear from those instructions
> if that macro is also used when building numpy.  I ran the build command
> twice, once before defining the variable and once after defining it, and it
> does change how numpy is built--well, it changes the names of some
> directories, anyway.  Is this macro also supposed to be defined when
> building numpy?

Don't bother for either numpy or scipy.

Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco

More information about the SciPy-user mailing list