[Python-bugs-list] [ python-Bugs-610730 ] Numpy doesn't build for Python.framework
noreply@sourceforge.net
noreply@sourceforge.net
Tue, 08 Oct 2002 01:54:09 -0700
Bugs item #610730, was opened at 2002-09-17 21:31
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=105470&aid=610730&group_id=5470
Category: Macintosh
>Group: 3rd Party
>Status: Closed
Resolution: None
Priority: 5
Submitted By: Just van Rossum (jvr)
Assigned to: Jack Jansen (jackjansen)
Summary: Numpy doesn't build for Python.framework
Initial Comment:
Running "python setup.py build" in a recent Numeric folder ends
with this error:
building 'umath' extension
gcc -DNDEBUG -g -O3 -Wall -Wstrict-prototypes -Wno-long-
double -no-cpp-precomp
-fno-common -dynamic -DHAVE_INVERSE_HYPERBOLIC -
IInclude -IPackages/FFT/Include
-IPackages/RNG/Include
-I/Library/Frameworks/Python.framework/Versions/2.3/include/
python2.3 -c
Src/umathmodule.c -o build/temp.darwin-5.5-Power Macintosh-2.3/
umathmodule.o
Src/umathmodule.c: In function `UINT_multiply':
Src/umathmodule.c:718: warning: `x' might be used uninitialized in
this function
Src/umathmodule.c: At top level:
Src/umathmodule.c:2394: `acosh' undeclared here (not in a
function)
Src/umathmodule.c:2394: too many errors, bailing out
error: command 'gcc' failed with exit status 1
[python:~/code/Numeric-22.0] just%
Seth Delackner reports:
"""I had this *exact* same problem until I nuked my old
/usr/local/bin/python* and reinstalled from 2.2.1's release
source.tar.gz, following letter for letter the instructions in the
Mac/OSX/README file. Thing to remember is that installing the
framework
build does NOT create simlinks in /usr/local/bin, so old versions of
python sitting there can screw it all up."""
I am not willing to do this until Python.framework is fully working,
so I can't verify this workaround...
----------------------------------------------------------------------
>Comment By: Jack Jansen (jackjansen)
Date: 2002-10-08 10:54
Message:
Logged In: YES
user_id=45365
Yes, it's a numpy bug (or, alternatively, a MacOSX bug). MacOSX doesn't cave the hyperbolic functions, so numpy should cater for that.
Can you pass the bug on to them, just to be sure? Probably Rob has already done that, but you never know...
----------------------------------------------------------------------
Comment By: Just van Rossum (jvr)
Date: 2002-10-08 01:43
Message:
Logged In: YES
user_id=92689
This does indeed fix the problem for me. Does this mean it's a NumPy
bug?
----------------------------------------------------------------------
Comment By: Jack Jansen (jackjansen)
Date: 2002-10-07 16:41
Message:
Logged In: YES
user_id=45365
Rob Managhan posted the following suggestion on pythonmac-sig, could you please try it?
Set and I were having problems compiling the Python Numeric module for OSX 10.1.5 sue to missing inverse hyperbolic functions.
I added these lines to setup.py (the first two are already there)
elif sys.platform in ['mac', 'beos5']:
mathlibs = []
elif sys.platform in ['darwin']:
undef_macros = ['HAVE_INVERSE_HYPERBOLIC']
----------------------------------------------------------------------
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=105470&aid=610730&group_id=5470