[SciPy-dev] Further Solaris woes

eric eric at scipy.org
Fri Feb 15 15:18:47 EST 2002


----- Original Message ----- 
From: "Tom Loredo" <loredo at astrosun.astro.cornell.edu>
To: <scipy-dev at scipy.org>
Sent: Friday, February 15, 2002 4:14 PM
Subject: [SciPy-dev] Further Solaris woes


> 
> Hi folks-
> 
> Thanks for the continuing Solaris help.  Sorry I didn't catch
> the FFTW install detail---I had already installed FFTW the
> recommended way for use with Travis's old wrappers for Python 2.1,
> and it worked fine that way.  The devil's in the details I guess.
> 
> Alex's patch fixes the pstat.py problem, but now I get an
> error when _quadpack.so is loaded:
> 
>     import _quadpack
> ImportError: ld.so.1: python: fatal: relocation error: file /home/laplace/lib/python2.2/site-packages/scipy/integrate/_quadpack.so: symbol __vlog_: referenced symbol not found
> 
> I'm guessing I need to add a library somewhere, but I'm not sure
> where to find __vlog_.  Any advice would be appreciated.  

My standard approach is the go in the library directory for the compiler and do an "nm" which preints out the symbols in the libraries and then grep for the one I want.

[15] eaj2 at teer4% cd /usr/pkg/spro/sun4m_55/SC4.0/lib
[14] eaj2 at teer4% nm *.a | grep vlog
libmvec.a[vlog_.o]:
[7]     |         0|       0|NOTY |GLOB |0    |UNDEF  |__vlog
[6]     |         0|      32|FUNC |GLOB |0    |2      |__vlog_
[5]     |         0|      32|FUNC |WEAK |0    |2      |vlog_
[1]     |         0|       0|FILE |LOCL |0    |ABS    |vlog_.c
[7]     |         0|       0|NOTY |GLOB |0    |UNDEF  |__vlog
libmvec.a[__vlog.o]:
[48]    |         0|    3844|FUNC |GLOB |0    |2      |__vlog
[1]     |         0|       0|FILE |LOCL |0    |ABS    |__vlog.s
libmvec_mt.a[vlog_.o]:
[13]    |         0|       0|NOTY |GLOB |0    |UNDEF  |__vlog
[11]    |       120|     200|FUNC |GLOB |0    |2      |__vlog_
[12]    |         0|     104|FUNC |GLOB |0    |2      |__vlog_mfunc
[10]    |       120|     200|FUNC |WEAK |0    |2      |vlog_
[1]     |         0|       0|FILE |LOCL |0    |ABS    |vlog_.c
[7]     |         0|       0|NOTY |GLOB |0    |UNDEF  |__vlog
libmvec_mt.a[__vlog.o]:
[48]    |         0|    3844|FUNC |GLOB |0    |2      |__vlog
[1]     |         0|       0|FILE |LOCL |0    |ABS    |__vlog.s

I'm not familiar with Sun's nm output, but it looks like some variation of your missing function is defined in on of the above.  I don't see a signature that matches __vlog__ exactly.  Sometimes there is an underscore mismatch that the linker handles internally (at least on windows with gcc that appears to be the case). You might try adding -mvec to the command and see if that solves the problem, but I'll bet there is an underscore issue here.

see ya,
eric

> The setup script ends up generating the following build command for
> _quadpack.so:
> 
> gcc -shared build/temp.solaris-2.7-sun4u-2.2/_quadpackmodule.o -L(null) -LSun/lib -L(null) -LSun/lib -Lbuild/temp.solaris-2.7-sun4u-2.2 -Lbuild/temp.solaris-2.7-sun4u-2.2 -Wl,-R(null) -Wl,-RSun/lib -lamos -ltoms -lfitpack -lminpack -lquadpack -lodepack -llinpack_lite -lblas -lmach -lF77 -lM77 -lsunmath -lm -lgist -lc_misc -lcephes -o build/lib.solaris-2.7-sun4u-2.2/scipy/integrate/_quadpack.so -mimpure-text
> 
> Thanks,
> Tom
> _______________________________________________
> Scipy-dev mailing list
> Scipy-dev at scipy.net
> http://www.scipy.net/mailman/listinfo/scipy-dev
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/scipy-dev/attachments/20020215/603e6eba/attachment.html>


More information about the SciPy-Dev mailing list