[FieldTrip] [multivariate] MEX compiling for 32-bit MATLAB under 64-bit Linux: who helps?

[Sander Maijers]

Jen, thanks for your suggestion. That problem I had fixed before. I am
among those who could not even get a 64-bit install if they wished so,
because the Student edition of MATLAB is still 32-bit-only.

This problem will occur later on, in case some MEX-compilation needs to
be done, in my case with the multivariate module!

I have succeeded in fixing the problem however. Those who are
confronted with it as well, can read:
http://fieldtrip.fcdonders.nl/faq/compile?&#fixing_mex-compilation_with_32-bit_variants_of_matlab_running_under_64-bit_linux .


On Wed, 25 May 2011 10:53:28 -0700
Jen Whitman <jenwhitman at gmail.com> wrote:

> As another ubuntu 64-bit user, I'd say your problem is with your
> matlab install rather than with fieldtrip. You may have inadvertently
> specified 32-bit instead of 64-bit when you installed matlab. If you
> type "computer" at the matlab prompt, it should say "GLNXA64". If
> not, I recommend redoing the matlab installation and making sure you
> specify your OS as 64-bit. If you still get error message about
> libstdc++.so, you may need to install that library using the
> 'synaptic package manger' (in your system menu under administration).
> A good resource for help if you have further difficulty may be
> linuxquestions.org.
> 
> Best of luck!
> 
> On Wed, May 25, 2011 at 2:50 AM, Sander Maijers
> <S.N.Maijers at student.ru.nl>wrote:
> 
> > Dear readers,
> >
> > I am running MATLAB R2011a 32-bit under Ubuntu Linux 10.10 64-bit
> > (Linux thuisrekenaar 2.6.35-28-generic #50-Ubuntu SMP Fri Mar 18
> > 18:42:20 UTC 2011 x86_64 GNU/Linux).
> >
> > When I call "ft_timelockstatistics()" with "cfg.mva
> > ={ft_mv_standardizer() ft_mv_svm()};"
> >
> > It fails:
> >
> > "selected 60 channels
> > selected 351 time bins
> > selected 1 frequency bins
> > using "statistics_crossvalidate" for the statistical testing
> > initializing random number generator with seed 1
> > validating 1 dataset(s)
> > input 1 consists of 439 examples and 21060 features
> > output 1 consists of 439 examples and 1 features
> > validating using 5-fold cross-validation
> > using 5 folds for 1 datasets
> > dataset 1: validating fold 1 of 5 using 351 training samples and 88
> > test samples Compiling tprod for first use
> > /usr/bin/ld: skipping
> > incompatible /usr/lib/gcc/x86_64-linux-gnu/4.3.5/libstdc++.so when
> > searching for -lstdc++ /usr/bin/ld: skipping
> > incompatible /usr/lib/gcc/x86_64-linux-gnu/4.3.5/libstdc++.a when
> > searching for -lstdc++ /usr/bin/ld: skipping
> > incompatible /usr/lib/gcc/x86_64-linux-gnu/4.3.5/libstdc++.so when
> > searching for -lstdc++ /usr/bin/ld: skipping
> > incompatible /usr/lib/gcc/x86_64-linux-gnu/4.3.5/libstdc++.a when
> > searching for -lstdc++ /usr/bin/ld:
> >        cannot find -lstdc++ collect2: ld returned 1 exit status
> >
> > mex: link of ' "tprod.mexglx"' failed.
> >
> > ??? Error using ==> tprod at 132 unable to compile MEX version of
> > '/mnt/data1/0/9/data/software/UNIX/scientific/FieldTrip/bin/fieldtrip-read-only/multivariate/external/farquhar/tprod/tprod',
> > please make sure your MEX compiler is set up correctly (try 'mex
> > -setup')."
> >
> >
> > I have followed the instructions on
> > https://help.ubuntu.com/community/MATLAB#MEX functions to enable MEX
> > compiling but it was not successful.
> >
> > -> Before I singlehandedly bring my operating system configuration
> > into disarray, I would like to learn from you the following. Do you
> > know of ways to use a gcc to compile code written for a 32-bit
> > architecture ona 64-bit machine? Or if necessary another compiler
> > that works under Linux? -> If such a thing is certainly impossible,
> > are there ways to avoid MEX compiling? Or could someone provide
> > precompiled MEX files that would suit my configuration?
> >
> > [*] I have installed the Ubuntu package "ia32-libs" (version
> > 20090808ubuntu9.1)
> > [*] It seems that the only thing I need is 32-bit binaries for
> > libstdc++. Perhaps I can introduce these in some creative way, but
> > I would like to check with you all first.
> >
> > I hope you are willing to help me. As a novice to FieldTrip and even
> > MATLAB, I have encountered many obstacles but I am determined to
> > surmount them.
> >
> > Kind regards,
> > Sander Maijers (BA Linguistics student)
> > _______________________________________________
> > fieldtrip mailing list
> > fieldtrip at donders.ru.nl
> > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> >