[SciPy-user] disabling fftw3 during scipy build

Robert Kern robert.kern@gmail....
Thu Feb 14 15:11:16 CST 2008

On Thu, Feb 14, 2008 at 2:48 PM, Russell E. Owen <rowen@cesmail.net> wrote:
> In article
>  <cd7634ce0802131632s427ad312ud6e705b202e0bcc1@mail.gmail.com>,
>  "Barry Wark" <barrywark@gmail.com> wrote:
>  > Hi all,
>  >
>  > I would be happy to start producing OS X eggs of scipy for
>  > distribution. At our site, we build scipy with fftw3 from MacPorts. I
>  > know the preference is to build eggs for release without the fftw3
>  > dependency. Is there a way to turn off linking with fftw3 during the
>  > scipy build process, even if fftw3 is present on the system? I can
>  > disable fftw3 via MacPorts, build, then re-enable, but I'd prefer to
>  > keep the build process a little more contained. Forgive me if rkern's
>  > already told us all how to do this years ago...
>  Would it make sense to statically link fftw3, instead? To do that I
>  think all you would have to do is delete (or temporarily hide) the
>  shared library, build scipy, then the shared library back. Then all
>  users of your egg would gain the benefits of fftw3.
>  I'm sure there is a more elegant solution involving copying the fftw3
>  static library in a special directory and using environment variable
>  magic to make it visible and fink NOT visible to the scipy build. But I
>  don't know the magic (if I did I'd use it to build matplotlib).

The reason FFTW is discouraged in this case is because it is GPLed.
For official/semi-official binaries, we are asking that no GPLed code
be included in the binary. We would like to be able to say that the
official binaries are under the BSD license. For unofficial binaries,
we have no opinion, of course.

In order to disable the use of FFTW, use the environment variable
"FFTW" like so:

  $ FFTW=None python setup.py build

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