help-octave
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Can't install octave-forge packages in a 32-bit OS X build, but can


From: Alexander Hansen
Subject: Re: Can't install octave-forge packages in a 32-bit OS X build, but can in a 64-bit one
Date: Sat, 05 Nov 2011 08:54:20 -0400
User-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.5; rv:7.0.1) Gecko/20110929 Thunderbird/7.0.1

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 11/4/11 9:03 PM, Alexander Hansen wrote:
> Hi.  I'm the maintainer for Octave for the Fink project,  serving
> Mac OS X.
> 
> One of our users reported a problem in Octave-3.4.3, built as
> 32-bit, on Mac OS 10.6.8, which I confirmed, wherein 'pkg' was
> unable to install anything, e.g :
> 
> octave:1> pkg -verbose install physicalconstants-0.1.7.tar.gz mkdir
> (/var/tmp/oct-2rjbOh) untar (physicalconstants-0.1.7.tar.gz,
> /var/tmp/oct-2rjbOh) invalid assignment to cs-list outside multiple
> assignment error: called from `dir' in file 
> /sw32/share/octave/3.4.3/m/miscellaneous/dir.m near line 115,
> column 26 error: called from: error:
> /sw32/share/octave/3.4.3/m/pkg/pkg.m at line 832, column 5 error:
> /sw32/share/octave/3.4.3/m/pkg/pkg.m at line 361, column 9
> 
> I've tried removing some semicolons to see if I can debug where
> this goes awry, but it appears to be simply the assignment on line
> 115 of dir.m:
> 
> info(i,1).name = fn;
> 
> 
> In contrast, a 64-bit build on the same machine works properly.
> The two builds use the same C and C++ compilers:
> 
> octave:4> octave_config_info('CC_VERSION') ans = 4.2.1 (Apple Inc.
> build 5666) (dot 3) octave:5> octave_config_info('CXX_VERSION') ans
> = 4.2.1
> 
> (Xcode tools 3.2.6) but different Fortran compilers:  a 32-bit 
> gfortran and a 64-bit gfortran, respectively, for the 32-bit and 
> 64-bit Octave.
> 
> Apart from the expected differences I'd naively expect from 32 vs.
> 64 bits, I also saw a few other things.  The 64-bit build's
> configuration had:
> 
> checking for working mktime... no
> 
> vs. "yes" on the 32-bit one.  Also, the 64-bit build had
> 
> checking for working nanosleep... no (mishandles large arguments)
> 
> whereas the 32-bit one had "yes".  Only the 32-bit build was given
> the -mieee-fp modifier ("flag-sort -r" is a Perl script that we use
> that changes the ordering of arguments to the compiler--it was
> needed in this case because of a conflict with another package):
> 
> checking whether flag-sort -r gcc accepts -mieee-fp... yes 
> configure: adding -mieee-fp to XTRA_CFLAGS checking whether
> flag-sort -r g++ accepts -mieee-fp... yes configure: adding
> -mieee-fp to XTRA_CXXFLAGS ... checking whether
> /sw32/bin/gfortran-fsf-4.6 accepts -mieee-fp... yes adding
> -mieee-fp to FFLAGS configure: defining FFLAGS to be -O3 -mieee-fp
> 
> 
> In the build output, the only thing that seemed to be different 
> between the two cases was the presence of the -mieee-fp flag.
> 
> I'm stumped here, so I'd appreciate any suggestions, and I'll
> gladly provide any additional information.

A couple of addenda:

1) Just to be clear:  the 32-bit and 64-bit builds use the same
configuration instructions on my end.

2) I see the same thing on OS X 10.5:  64-bit works properly, 32-bit
doesn't.

3) The user who reported the problem to me has a simple test case to
illustrate the failure:

On my 64-bit build:

octave-3.4.3:1> info = struct (zeros (0, 1), {"name", "date", "bytes",
"isdir","datenum", "statinfo"});
octave-3.4.3:2> info(1,1).name = "asdf"
info =

  scalar structure containing the fields:

    name = asdf
    date = [](0x0)
    bytes = [](0x0)
    isdir = [](0x0)
    datenum = [](0x0)
    statinfo = [](0x0)

On the 32-bit build:

octave-3.4.3:1> info = struct (zeros (0, 1), {"name", "date", "bytes",
"isdir","datenum", "statinfo"});
octave-3.4.3:2> info(1,1).name = "asdf"
error: invalid assignment to cs-list outside multiple assignment
- -- 
Alexander Hansen
Fink User Liaison
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk61MfwACgkQB8UpO3rKjQ9M1wCfWxKiDmBL5/Y+EXOx9wdL9KLk
aVQAn2rkNKsJ+9566dpDAdQU9kapO9cL
=3ArW
-----END PGP SIGNATURE-----


reply via email to

[Prev in Thread] Current Thread [Next in Thread]