Re: Installing on OSX i386: invalid assignment to cs-list outside multiple assignment

[Alexander Hansen]

On 4/17/12 4:26 PM, Joel Nothman wrote:
> On Wed, 18 Apr 2012 03:02:25 +1000, Alexander Hansen 
> <address@hidden> wrote:
>
> > On 4/17/12 12:49 AM, Joel Nothman wrote:
> > > Hi,
> > >
> > > An Octave-help thread in November was titled 'Can't install 
> > > octave-forge packages in a 32-bit OS X build, but can in a 64-bit one'.
> > >
> > > It presented an issue where the 'invalid assignment to cs-list 
> > > outside multiple assignment' error occurred where it should not.
> > >
> > > For example:
> > > > x = [];
> > > > x(1).a1 = 1;
> > > > x(2).a1 = 2;
> > > > x(1).a2 = 3;
> > > error: ...
> > >
> > > It was resolved for Fink's octave-3.6.1.info by using CXXFLAGS=-O0 
> > > when compiling Octave 3.4.3 on OS X 10.5 (32 bit).
> > >
> > >
> > > I am experiencing the same problem in Octave 3.4.3 and 3.6.1, but I 
> > > am attempting to use Homebrew rather than Fink.
> > >
> > > I have tried adding
> > >   ENV['CXXFLAGS'] = '-O0' if !ENV.m64
> > > prior to system calls to configure and make (at 
> > > https://github.com/mxcl/homebrew/blob/master/Library/Formula/octave.rb:73).
> > >
> > > This results in:
> > >
> > > configure: error: your Fortran compiler must have an option to make 
> > > integers the same size as octave_idx_type (int).  See the file 
> > > INSTALL for more information.
> > >
> > > I can't find any relevant information in INSTALL (but will admit 
> > > I've not read its every word).
> > >
> > > Could you please help me getting this compiled and operating correctly?
> > >
> > > Thanks,
> > >
> > > - Joel
> > I'd check the config.log file, too.
>
>
> Yes, you're probably right. The section where the error occurs (Octave 
> 3.6.1):
>
> configure:34402: checking whether /usr/local/bin/gfortran generates 
> correct size integers
> configure:34417: /usr/local/bin/gfortran -c -Os -w -pipe   
> -march=core2 -msse4.1 -mieee-fp  conftest.f >&5
> configure:34417: $? = 0
> configure:34466: /usr/bin/gcc-4.2 -o conftest -Os -w -pipe 
> -march=core2 -msse4.1 -D_REENTRANT -m64 -D_THREAD_SAFE -pthread 
> -I/usr/local/Cellar/curl/7.24.0/include 
> -I/usr/local/Cellar/readline/6.2.2/include 
> -I/usr/local/Cellar/texinfo/4.13a/include -I/usr/X11/include 
> -L/usr/local/Cellar/curl/7.24.0/lib 
> -L/usr/local/Cellar/readline/6.2.2/lib 
> -L/usr/local/Cellar/texinfo/4.13a/lib -L/usr/X11/lib -arch x86_64 
> conftest.c fintsize.o  -L/usr/local/Cellar/curl/7.24.0/lib 
> -L/usr/local/Cellar/readline/6.2.2/lib 
> -L/usr/local/Cellar/texinfo/4.13a/lib -L/usr/X11/lib 
> -L/usr/local/Cellar/gfortran/4.2.4-5577/bin/../lib/gcc/i686-apple-darwin9/4.2.1/x86_64 
> -L/usr/lib/gcc/i686-apple-darwin9/4.2.1/x86_64 
> -L/usr/lib/i686-apple-darwin9/4.2.1 
> -L/usr/local/Cellar/gfortran/4.2.4-5577/bin/../lib/gcc/i686-apple-darwin9/4.2.1 
> -L/usr/local/Cellar/gfortran/4.2.4-5577/bin/../lib/gcc 
> -L/usr/lib/gcc/i686-apple-darwin9/4.2.1 
> -L/usr/local/Cellar/gfortran/4.2.4-5577/bin/../lib/gcc/i686-apple-darwin9/4.2.1/../../.. 
> -L/usr/lib/gcc/i686-apple-darwin9/4.2.1/../../../i686-apple-darwin9/4.2.1 
> -L/usr/lib/gcc/i686-apple-darwin9/4.2.1/../../.. -lm -lgfortranbegin 
> -lgfortran >&5
> conftest.c: In function 'main':
> conftest.c:529: error: 'foo' undeclared (first use in this function)
> conftest.c:529: error: (Each undeclared identifier is reported only once
> conftest.c:529: error: for each function it appears in.)
> conftest.c:529: error: 'FOO' undeclared (first use in this function)
> conftest.c:529: error: called object 'F77_FUNC(<erroneous-expression>, 
> <erroneous-expression>)' is not a function
> configure:34466: $? = 1
> configure: program exited with status 1
>
> So the error is not an assertion failure. Why would it break if 
> CXXFLAGS='-O0'?
>
> Thanks,
>
> - Joel
No clue if they're connected.  I don't see a '-O0' in the above.  
However, that -m64 looks a bit suspicious in a 32-bit build.

--
Alexander Hansen, Ph.D.
Fink User Liaison
http://finkakh.wordpress.com/2012/02/21/got-job/