Compile fails for r3123 on OS X 10.4 (Intel)
I've been unable to compile scipy on OS X 10.4 (intel) from the recent trunk. Scipy built on this machine as of r2708. The output from python setup.py build is attached. I have fftw3 installed via macports (in /opt/local), and it appears that the build finds it properly, but the build fails with an error: building extension "scipy.fftpack._fftpack" sources target build/src.macosx-10.3-fat-2.5/_fftpackmodule.c does not exist: Assuming _fftpackmodule.c was generated with "build_src --inplace" command. error: '_fftpackmodule.c' missing I would appreciate any advice or suggestions! Thanks, Barry
I've attached the output to ticket #449 on the scipy trac, which appears to have reported the same problem. On 6/29/07, Barry Wark <barrywark@gmail.com> wrote:
I've been unable to compile scipy on OS X 10.4 (intel) from the recent trunk. Scipy built on this machine as of r2708. The output from
python setup.py build
is attached.
I have fftw3 installed via macports (in /opt/local), and it appears that the build finds it properly, but the build fails with an error:
building extension "scipy.fftpack._fftpack" sources target build/src.macosx-10.3-fat-2.5/_fftpackmodule.c does not exist: Assuming _fftpackmodule.c was generated with "build_src --inplace" command. error: '_fftpackmodule.c' missing
I would appreciate any advice or suggestions!
Thanks, Barry
I know this isn't anyone's top priority on the list, but I'd really appreciate a pointer (or a dope slap): I've successfully compiled previous revisions (<r3000) of scipy on this system (OSX Intel) before. However, after checking out an older version, the build fails with File "Lib/odr/setup.py", line 9, in <module> from numpy.distutils.misc_util import get_path, Configuration, dot_join ImportError: cannot import name get_path I assume this is because numpy has been updated and removed get_path. Besides ticket #449, I haven't been able to find mention of the problem described earlier in the thread in the list archives or online. Obviously, not many other people are having this problem, so I assume it's a stupid mistake on my part. It's holding up deployment of software to our lab, however, so I would be eternally grateful for any help. Thanks, Barry On 7/1/07, Barry Wark <barrywark@gmail.com> wrote:
I've attached the output to ticket #449 on the scipy trac, which appears to have reported the same problem.
On 6/29/07, Barry Wark <barrywark@gmail.com> wrote:
I've been unable to compile scipy on OS X 10.4 (intel) from the recent trunk. Scipy built on this machine as of r2708. The output from
python setup.py build
is attached.
I have fftw3 installed via macports (in /opt/local), and it appears that the build finds it properly, but the build fails with an error:
building extension "scipy.fftpack._fftpack" sources target build/src.macosx-10.3-fat-2.5/_fftpackmodule.c does not exist: Assuming _fftpackmodule.c was generated with "build_src --inplace" command. error: '_fftpackmodule.c' missing
I would appreciate any advice or suggestions!
Thanks, Barry
Barry Wark wrote:
I know this isn't anyone's top priority on the list, but I'd really appreciate a pointer (or a dope slap):
I've successfully compiled previous revisions (<r3000) of scipy on this system (OSX Intel) before. However, after checking out an older version, the build fails with
File "Lib/odr/setup.py", line 9, in <module> from numpy.distutils.misc_util import get_path, Configuration, dot_join ImportError: cannot import name get_path
I assume this is because numpy has been updated and removed get_path.
Besides ticket #449, I haven't been able to find mention of the problem described earlier in the thread in the list archives or online. Obviously, not many other people are having this problem, so I assume it's a stupid mistake on my part. It's holding up deployment of software to our lab, however, so I would be eternally grateful for any help.
No, it was a stupid mistake on my part. I needed to refactor that bit of code in numpy.distutils, and I thought the function was only used internally. The current SVN of scipy/odr/setup.py uses the correct idiom for building with numpy.distutils. While I shouldn't have removed get_path(), it is a holdover from the days of scipy_distutils. -- 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
On Fri, Jun 29, 2007 at 10:33:24AM -0700, Barry Wark wrote:
I've been unable to compile scipy on OS X 10.4 (intel) from the recent trunk. Scipy built on this machine as of r2708. The output from
python setup.py build
is attached.
I have fftw3 installed via macports (in /opt/local), and it appears that the build finds it properly, but the build fails with an error:
building extension "scipy.fftpack._fftpack" sources target build/src.macosx-10.3-fat-2.5/_fftpackmodule.c does not exist: Assuming _fftpackmodule.c was generated with "build_src --inplace" command. error: '_fftpackmodule.c' missing
I would appreciate any advice or suggestions!
I've got pretty much the same setup, and no problems. Did you delete the build directory first (rm -rf build)? -- |>|\/|< /--------------------------------------------------------------------------\ |David M. Cooke http://arbutus.physics.mcmaster.ca/dmc/ |cookedm@physics.mcmaster.ca
Yes, I'm using a clean checkout from SVN. Just to double check against your setup, I'm using fftw-3 installed via macports as my fftw library Barry On 7/3/07, David M. Cooke <cookedm@physics.mcmaster.ca> wrote:
On Fri, Jun 29, 2007 at 10:33:24AM -0700, Barry Wark wrote:
I've been unable to compile scipy on OS X 10.4 (intel) from the recent trunk. Scipy built on this machine as of r2708. The output from
python setup.py build
is attached.
I have fftw3 installed via macports (in /opt/local), and it appears that the build finds it properly, but the build fails with an error:
building extension "scipy.fftpack._fftpack" sources target build/src.macosx-10.3-fat-2.5/_fftpackmodule.c does not exist: Assuming _fftpackmodule.c was generated with "build_src --inplace" command. error: '_fftpackmodule.c' missing
I would appreciate any advice or suggestions!
I've got pretty much the same setup, and no problems. Did you delete the build directory first (rm -rf build)?
-- |>|\/|< /--------------------------------------------------------------------------\ |David M. Cooke http://arbutus.physics.mcmaster.ca/dmc/ |cookedm@physics.mcmaster.ca _______________________________________________ SciPy-user mailing list SciPy-user@scipy.org http://projects.scipy.org/mailman/listinfo/scipy-user
On Tue, Jul 03, 2007 at 01:14:12PM -0700, Barry Wark wrote:
Yes, I'm using a clean checkout from SVN. Just to double check against your setup, I'm using fftw-3 installed via macports as my fftw library
Barry
Still don't have your problem. Scipy is r3143, numpy is r3880. OS X 10.4.10 (Intel MacBook), Python 2.5 and 2.4. I'm using gfortran from macports, but that shouldn't affect this. One thing I can think of is checking your numpy install, as the _fortranmodule.c is generated by f2py. (in general, the latest svn scipy usually needs a recent svn numpy)
On 7/3/07, David M. Cooke <cookedm@physics.mcmaster.ca> wrote:
On Fri, Jun 29, 2007 at 10:33:24AM -0700, Barry Wark wrote:
I've been unable to compile scipy on OS X 10.4 (intel) from the recent trunk. Scipy built on this machine as of r2708. The output from
python setup.py build
is attached.
I have fftw3 installed via macports (in /opt/local), and it appears that the build finds it properly, but the build fails with an error:
building extension "scipy.fftpack._fftpack" sources target build/src.macosx-10.3-fat-2.5/_fftpackmodule.c does not exist: Assuming _fftpackmodule.c was generated with "build_src --inplace" command. error: '_fftpackmodule.c' missing
I would appreciate any advice or suggestions!
I've got pretty much the same setup, and no problems. Did you delete the build directory first (rm -rf build)?
-- |>|\/|< /--------------------------------------------------------------------------\ |David M. Cooke http://arbutus.physics.mcmaster.ca/dmc/ |cookedm@physics.mcmaster.ca _______________________________________________ SciPy-user mailing list SciPy-user@scipy.org http://projects.scipy.org/mailman/listinfo/scipy-user
_______________________________________________ SciPy-user mailing list SciPy-user@scipy.org http://projects.scipy.org/mailman/listinfo/scipy-user
-- |>|\/|< /--------------------------------------------------------------------------\ |David M. Cooke http://arbutus.physics.mcmaster.ca/dmc/ |cookedm@physics.mcmaster.ca
OK. This is a weird one for me... numpy r3881 builds fine. Scipy r3145 (a fresh checkout) does not build still. It looks like the scipy setup/config can't find many of the sources in sub-packages. Most relevant may be these lines from the build log: non-existing path in 'scipy/fftpack': 'tests' could not resolve pattern in 'scipy/fftpack': 'dfftpack/*.f' non-existing path in 'scipy/fftpack': 'fftpack.pyf' non-existing path in 'scipy/fftpack': 'src/zfft.c' non-existing path in 'scipy/fftpack': 'src/drfft.c' non-existing path in 'scipy/fftpack': 'src/zrfft.c' non-existing path in 'scipy/fftpack': 'src/zfftnd.c' non-existing path in 'scipy/fftpack': 'src/zfft_djbfft.c' non-existing path in 'scipy/fftpack': 'src/zfft_fftpack.c' non-existing path in 'scipy/fftpack': 'src/zfft_fftw.c' non-existing path in 'scipy/fftpack': 'src/zfft_fftw3.c' non-existing path in 'scipy/fftpack': 'src/zfft_mkl.c' non-existing path in 'scipy/fftpack': 'convolve.pyf' non-existing path in 'scipy/fftpack': 'src/convolve.c' Similar lines for many other packages are printed as well. Obviously none of these lines appear on a machine that does build scipy correctly. Can any of the numpy.distutils gurus point me in a helpful direction? Thanks! Barry On 7/3/07, David M. Cooke <cookedm@physics.mcmaster.ca> wrote:
On Tue, Jul 03, 2007 at 01:14:12PM -0700, Barry Wark wrote:
Yes, I'm using a clean checkout from SVN. Just to double check against your setup, I'm using fftw-3 installed via macports as my fftw library
Barry
Still don't have your problem. Scipy is r3143, numpy is r3880. OS X 10.4.10 (Intel MacBook), Python 2.5 and 2.4. I'm using gfortran from macports, but that shouldn't affect this.
One thing I can think of is checking your numpy install, as the _fortranmodule.c is generated by f2py.
(in general, the latest svn scipy usually needs a recent svn numpy)
On 7/3/07, David M. Cooke <cookedm@physics.mcmaster.ca> wrote:
On Fri, Jun 29, 2007 at 10:33:24AM -0700, Barry Wark wrote:
I've been unable to compile scipy on OS X 10.4 (intel) from the recent trunk. Scipy built on this machine as of r2708. The output from
python setup.py build
is attached.
I have fftw3 installed via macports (in /opt/local), and it appears that the build finds it properly, but the build fails with an error:
building extension "scipy.fftpack._fftpack" sources target build/src.macosx-10.3-fat-2.5/_fftpackmodule.c does not exist: Assuming _fftpackmodule.c was generated with "build_src --inplace" command. error: '_fftpackmodule.c' missing
I would appreciate any advice or suggestions!
I've got pretty much the same setup, and no problems. Did you delete the build directory first (rm -rf build)?
-- |>|\/|< /--------------------------------------------------------------------------\ |David M. Cooke http://arbutus.physics.mcmaster.ca/dmc/ |cookedm@physics.mcmaster.ca _______________________________________________ SciPy-user mailing list SciPy-user@scipy.org http://projects.scipy.org/mailman/listinfo/scipy-user
_______________________________________________ SciPy-user mailing list SciPy-user@scipy.org http://projects.scipy.org/mailman/listinfo/scipy-user
-- |>|\/|< /--------------------------------------------------------------------------\ |David M. Cooke http://arbutus.physics.mcmaster.ca/dmc/ |cookedm@physics.mcmaster.ca _______________________________________________ SciPy-user mailing list SciPy-user@scipy.org http://projects.scipy.org/mailman/listinfo/scipy-user
participants (3)
-
Barry Wark -
David M. Cooke -
Robert Kern