Mailman 3 April 2000 - Distutils-SIG

distutils+numpy problem on AIX
by Mike Hammill April 18, 2000

April 18, 2000

Hello, Does anyone know why I might get the following messages when trying to use distutils 0.8 to install NumPy on an AIX machine? unable to execute ./ld_so_aix: No such file or directory error: command './ld_so_aix' failed with exit status 1 Background: My goal is to get Python 1.5.2 running with NumPy for the users of our SP at the Royal Institute of Technology. I recently downloaded NumPy from LLNL: LLNLDistribution11.tgz and Numerical-14.tar.gz According to <ftp://ftp-icf.llnl.gov/pub/python/README.html> one is supposed to replace the Numerical part of the 11 distribution with the 14 one. I did this and then went to the Numerical subdirectory to install it. Following the suggestion in the README.txt in the distribution of Distutils-0.8, I replaced the setup.py file in Numerical with the one that comes in the examples directory of Distutils-0.8. When I ran python (1.5.2) with this, I get the messages below. Does anyone have any idea what might be going wrong? Notes: Python 1.5.2 seems to test fine on our SP (AIX 4.3). I installed it in a non-standard place (my directory) for testing. python setup.py install running install running build running build_py not copying Lib/ArrayPrinter.py (output up-to-date) not copying Lib/FFT.py (output up-to-date) not copying Lib/LinearAlgebra.py (output up-to-date) not copying Lib/MLab.py (output up-to-date) not copying Lib/Matrix.py (output up-to-date) not copying Lib/Numeric.py (output up-to-date) not copying Lib/Precision.py (output up-to-date) not copying Lib/RandomArray.py (output up-to-date) not copying Lib/UserArray.py (output up-to-date) running build_ext building '_numpy' extension skipping Src/_numpymodule.c (build/temp.aix3-002005909400/Src/_numpymodule.o up-to-date) skipping Src/arrayobject.c (build/temp.aix3-002005909400/Src/arrayobject.o up-to-date) skipping Src/ufuncobject.c (build/temp.aix3-002005909400/Src/ufuncobject.o up-to-date) ./ld_so_aix cc build/temp.aix3-002005909400/Src/_numpymodule.o build/temp.aix3-002005909400/Src/arrayobject.o build/temp.aix3-002005909400/Src /ufuncobject.o -o build/lib.aix3-002005909400/_numpy.so unable to execute ./ld_so_aix: No such file or directory error: command './ld_so_aix' failed with exit status 1 By the way, after this failure, I tried to just install NumPy strickly from LLNL's "11" distribution, using makethis.py. I get a different, but similar message, python makethis.py Listing . ... Listing ./Demo ... Compiling ./Demo/life.py ... Compiling ./Demo/mandelbrot.py ... Compiling ./Demo/sieve.py ... Listing ./Doc ... Listing ./Include ... Listing ./Lib ... Compiling ./Lib/ArrayPrinter.py ... Compiling ./Lib/FFT.py ... Compiling ./Lib/LinearAlgebra.py ... Compiling ./Lib/MLab.py ... Compiling ./Lib/Matrix.py ... Compiling ./Lib/Numeric.py ... Compiling ./Lib/Precision.py ... Compiling ./Lib/RandomArray.py ... Compiling ./Lib/UserArray.py ... Listing ./Src ... Listing ./Test ... Compiling ./Test/test_all.py ... Compiling ./installthis.py ... Compiling ./makethis.py ... rm -f *.o *~ rm -f *.a tags TAGS config.c Makefile.pre python sedscript rm -f *.so *.sl so_locations VERSION=`python -c "import sys; print sys.version[:3]"`; \ installdir=`python -c "import sys; print sys.prefix"`; \ exec_installdir=`python -c "import sys; print sys.exec_prefix"`; \ make -f ./Makefile.pre.in VPATH=. srcdir=. \ VERSION=$VERSION \ installdir=$installdir \ exec_installdir=$exec_installdir \ Makefile make[1]: Entering directory `/afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/p ymore/numpy/LLNLDistribution11/Numer sed -n \ -e '1s/.*/1i\\/p' \ -e '2s%.*%# Generated automatically from Makefile.pre.in by sedscript.%p' \ -e '/^VERSION=/s/^VERSION=[ ]*$.*$/s%@VERSION[@]%\1%/p' \ -e '/^CC=/s/^CC=[ ]*$.*$/s%@CC[@]%\1%/p' \ -e '/^CCC=/s/^CCC=[ ]*$.*$/s%#@SET_CCC[@]%CCC=\1%/p' \ -e '/^LINKCC=/s/^LINKCC=[ ]*$.*$/s%@LINKCC[@]%\1%/p' \ -e '/^OPT=/s/^OPT=[ ]*$.*$/s%@OPT[@]%\1%/p' \ -e '/^LDFLAGS=/s/^LDFLAGS=[ ]*$.*$/s%@LDFLAGS[@]%\1%/p' \ -e '/^LDLAST=/s/^LDLAST=[ ]*$.*$/s%@LDLAST[@]%\1%/p' \ -e '/^DEFS=/s/^DEFS=[ ]*$.*$/s%@DEFS[@]%\1%/p' \ -e '/^LIBS=/s/^LIBS=[ ]*$.*$/s%@LIBS[@]%\1%/p' \ -e '/^LIBM=/s/^LIBM=[ ]*$.*$/s%@LIBM[@]%\1%/p' \ -e '/^LIBC=/s/^LIBC=[ ]*$.*$/s%@LIBC[@]%\1%/p' \ -e '/^RANLIB=/s/^RANLIB=[ ]*$.*$/s%@RANLIB[@]%\1%/p' \ -e '/^MACHDEP=/s/^MACHDEP=[ ]*$.*$/s%@MACHDEP[@]%\1%/p' \ -e '/^SO=/s/^SO=[ ]*$.*$/s%@SO[@]%\1%/p' \ -e '/^LDSHARED=/s/^LDSHARED=[ ]*$.*$/s%@LDSHARED[@]%\1%/p' \ -e '/^CCSHARED=/s/^CCSHARED=[ ]*$.*$/s%@CCSHARED[@]%\1%/p' \ -e '/^SGI_ABI=/s/^SGI_ABI=[ ]*$.*$/s%@SGI_ABI[@]%\1%/p' \ -e '/^LINKFORSHARED=/s/^LINKFORSHARED=[ ]*$.*$/s%@LINKFORSHARED[@]%\1 %/p' \ -e '/^prefix=/s/^prefix=$.*$/s%^prefix=.*%prefix=\1%/p' \ -e '/^exec_prefix=/s/^exec_prefix=$.*$/s%^exec_prefix=.*%exec_prefix=\1%/p' \ /afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/python/lib/python1.5/config/M akefile >sedscript echo "/^#@SET_CCC@/d" >>sedscript echo "/^installdir=/s%=.*%= /afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix4 3/python%" >>sedscript echo "/^exec_installdir=/s%=.*%=/afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix4 3/python%" >>sedscript echo "/^srcdir=/s%=.*%= .%" >>sedscript echo "/^VPATH=/s%=.*%= .%" >>sedscript echo "/^LINKPATH=/s%=.*%= %" >>sedscript echo "/^BASELIB=/s%=.*%= %" >>sedscript echo "/^BASESETUP=/s%=.*%= %" >>sedscript sed -f sedscript ./Makefile.pre.in >Makefile.pre /afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/python/lib/python1.5/config/ma kesetup \ -m Makefile.pre -c /afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/py thon/lib/python1.5/config/config.c.in /afs/pdc.kth.se/home/m/mike/Python/.bin/ rs_aix43/python/lib/python1.5/config/Setup.thread /afs/pdc.kth.se/home/mhon/.bi n/rs_aix43/python/lib/python1.5/config/Setup.local /afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/python/l1.5/config/Setup make -f Makefile do-it-again make[2]: Entering directory `/afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/p ymore/numpy/LLNLDistribution11/Numer /afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/python/lib/python1.5/config/ma kesetup \ -m Makefile.pre -c /afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/py thon/lib/python1.5/config/config.c.in /afs/pdc.kth.se/home/m/mike/Python/.bin/ rs_aix43/python/lib/python1.5/config/Setup.thread /afs/pdc.kth.se/home/mhon/.bi n/rs_aix43/python/lib/python1.5/config/Setup.local /afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/python/l1.5/config/Setup make[2]: Leaving directory `/afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/py more/numpy/LLNLDistribution11/Numeri make[1]: Leaving directory `/afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/py more/numpy/LLNLDistribution11/Numeri cc -I./Include -O -I/afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/python/i nclude/python1.5 -I/afs/pdc.kth.se/he/Python/.bin/rs_aix43/python/include/pytho n1.5 -DHAVE_CONFIG_H -c ./Src/_numpymodule.c cc -I./Include -O -I/afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/python/i nclude/python1.5 -I/afs/pdc.kth.se/he/Python/.bin/rs_aix43/python/include/pytho n1.5 -DHAVE_CONFIG_H -c ./Src/arrayobject.c "./Src/arrayobject.c", line 1539.10: 1506-342 (W) "/*" detected in comment. cc -I./Include -O -I/afs/pdc.kth.se/home/m/mike/Python/.bin/rs_aix43/python/i nclude/python1.5 -I/afs/pdc.kth.se/he/Python/.bin/rs_aix43/python/include/pytho n1.5 -DHAVE_CONFIG_H -c ./Src/ufuncobject.c ./ld_so_aix cc _numpymodule.o arrayobject.o ufuncobject.o -o _numpymodule.so make: ./ld_so_aix: Command not found make: *** [_numpymodule.so] Error 127 Many thanks, Mike Hammill <mike(a)pdc.kth.se>

1 0

Alternate email address
by Greg Ward April 17, 2000

April 17, 2000

Anyone who has been trying to send me mail since last Friday: please resend it to gward(a)ase.com; as long as python.net is down, this will be my substitute address. Greg -- Greg Ward - Linux weenie gward(a)ase.com http://starship.python.net/~gward/ If you're not part of the solution, you're part of the precipitate.

1 0

Re: [Distutils] Patch for build_ext.py (Win32)
by Lyle Johnson April 14, 2000

April 14, 2000

> No, it's the job of the CCompiler classes to ensure that the files they > are about to write can be written, ie. that os.path.dirname(output_file) > exists before running the compile/link/whatever that attempts to create > output_file. > Yes, the problem is that during the link (i.e. when we're actually making the Windows DLL) it actually creates (well, tries to create) the following *three* files: build\lib.win32\MyExtensionName.pyd build\temp.win32\Release\MyExtensionName.lib build\temp.win32\Release\MyExtensionName.exp For the record, here's what my distutils-conjured link line looks like: link.exe /DLL /nologo /INCREMENTAL:NO /LIBPATH:../dtf \ /LIBPATH:D:\Progra~1\Python\libs DTF.lib \ build\temp.win32\Release\core_wrap.obj \ /OUT:build\lib.win32\dtf\core.pyd /export:initcore \ /IMPLIB:build\temp.win32\Release\dtf\core.lib You are correct that link_shared_object() takes care of mkpath'ing the output directory for the DLL itself but it does *not* create the needed directories for the scratch files ("core.lib" and "core.exp"). So perhaps the patch that I originally suggested for build_ext.py should be moved into link_shared_object() instead. It doesn't matter to me as long as somebody creates that directory before we try to dump files into it ;) Lyle

1 0

Possible fix for sysconfig.py (Win32)
by Lyle Johnson April 14, 2000

April 14, 2000

The call to parse_config_h() in _init_nt() [both found in distutils/sysconfig.py] clobbers the previously correct values of PREFIX and EXEC_PREFIX for the standard Python installation on Win32. This in turn leads to the compiler picking up the wrong Python include file paths (and probably other glitches I haven't stumbled onto yet). One workaround (the one I've used) is to comment out that call to parse_config_h() ;)

3 2

Patch for msvccompiler.py (Win32)
by Lyle Johnson April 13, 2000

April 13, 2000

Here's a patch for distutils/msvccompiler.py to ensure that the appropriate Python import library (e.g. "python15.lib" or "python15_d.lib") is linked in with your extension DLL on Windows. As my comment indicates, there's probably a cleaner way to pick up the correct library name but this should work for Python 1.5.x and Python 1.6. Index: msvccompiler.py =================================================================== RCS file: /projects/cvsroot/distutils/distutils/msvccompiler.py,v retrieving revision 1.25 diff -c -r1.25 msvccompiler.py *** msvccompiler.py 2000/03/31 19:04:25 1.25 --- msvccompiler.py 2000/04/12 22:43:28 *************** *** 329,334 **** --- 329,346 ---- extra_preargs=None, extra_postargs=None): + # Add the Python library to the standard library list. + # There's probably a more elegant way to determine the correct name ;) + if sys.winver == '1.5': + pythonlib = 'python15' + else: + pythonlib = 'python16' + + if debug: + pythonlib = pythonlib + '_d' + + self.add_library(pythonlib) + (objects, output_dir) = self._fix_object_args (objects, output_dir) (libraries, library_dirs, runtime_library_dirs) = \ self._fix_lib_args (libraries, library_dirs, runtime_library_dirs)

4 4

Patch for build_ext.py (Win32)
by Lyle Johnson April 13, 2000

April 13, 2000

Here's a patch for distutils/command/build_ext.py to ensure that the output directory for scratch files created while building the extension DLL actually exists. Index: build_ext.py =================================================================== RCS file: /projects/cvsroot/distutils/distutils/command/build_ext.py,v retrieving revision 1.30 diff -c -r1.30 build_ext.py *** build_ext.py 2000/04/10 00:19:42 1.30 --- build_ext.py 2000/04/12 22:33:10 *************** *** 331,336 **** --- 331,342 ---- implib_dir = os.path.join(self.build_temp,\ self.get_ext_libname(extension_name)) extra_args.append ('/IMPLIB:' + implib_dir) + + # Ensure that this directory exists! + ext_path = string.split(extension_name, '.') + ext_path = apply(os.path.join, ext_path[:-1]) + self.mkpath(os.path.join(self.build_temp, ext_path)) + # if MSVC fullname = self.get_ext_fullname (extension_name)

2 1

problem with installing NumPy under WinNT
by Berthold Hoellmann April 13, 2000

April 13, 2000

Hello, I tried to install NumPy under NT using setup_numpy.py from distutils 0.8. Compiling failed because Python.h could not be found. I tracked it down to the following: Python 1.5.2 (#0, Apr 13 1999, 10:51:12) [MSC 32 bit (Intel)] on win32 Copyright 1991-1995 Stichting Mathematisch Centrum, Amsterdam >>> from distutils import sysconfig >>> sysconfig.exec_prefix '""' >>> import sys >>> sys.exec_prefix 'G:\\bin\\Python' >>> sysconfig.EXEC_PREFIX '""' >>> I guess, sysconfig.EXEC_PREFIX should return '"G:\\bin\\Python"', but I have no idea, what's going wrong here. Thanks Berthold -- email: hoel(a)GermanLloyd.org ) ( C[_] These opinions might be mine, but never those of my employer.

1 0

Bug in distutils 0.8
by Berthold Hoellmann April 13, 2000

April 13, 2000

Hello, I guess i've found a small bug in distutils 0.8. sample3/setup.py shows how to extend the build command. Besides the missing function 'run_peer' in the Build class, it tries to load Build from command.build. I guess it should import command.build.build instead Greetings Berthold -- email: hoel(a)GermanLloyd.org ) ( C[_] These opinions might be mine, but never those of my employer.

1 0

problem when installing under WinNT
by Berthold Hoellmann April 13, 2000

April 13, 2000

Hello, I've written a package called qpPy, which I would like to install under WinNT. But when I import the successfull installed I get: ... NameError: Case mismatch for module name qpPy.py ... Is there any idea how to solve this (besides renaming the file by hand)? Thanks Berthold -- email: hoel(a)GermanLloyd.org ) ( C[_] These opinions might be mine, but never those of my employer.

1 0

Voodoo Python
by Lyle Johnson April 13, 2000

April 13, 2000

Please disregard the patch for msvccompiler.py which I submitted yesterday. After Thomas' e-mail I went back to try to reproduce the problem with the (unpatched) version of msvccompiler.py and it worked like a charm. Maybe I just needed a good night's sleep.

1 0