Re: MacOS X 10.3.9 build but load-save.m test failure

John W. Eaton wrote:
> On 27-Apr-2005, Bill Northcott <w.northcott@xxxxxxxxxxx> wrote:
>
> | I have succeded in building Octave 2.9.2 code on MacOS X 10.3.9
> | 
> | The C/C++ compiler was a prerelease gcc-4.00 built from code with the 
> | cvs tag apple-gcc-5009.
> | The gfortran in that compiler will not handle alternate entry points so 
> | g77 from hpc.sourceforge.net was used for the Fortran codes.
> | 
> | This results in the restFP, saveFP problem at link time as described at 
> | http://www.astro.gla.ac.uk/users/norman/note/2004/restFP/.  It would be 
> | good if this could be fixed in the autoconf stuff.   Re-running the 
> | links with the addition of '-lcc_dynamic' allowed the octave build to 
> | complete.
>
> It would be helpful if someone could provide a patch because I do not
> have access to an OS X system that I can use for testing.
>
> | The test suite run with with make check produced:
> | ...
> | Running ../../test/octave.test/io/io.exp ...
> | FAIL: octave.test/io/load-save.m
>
> Can you please investigate what is causing this test to fail?

For one thing, which optimization flags did you use?
I compiled 2.9.2 with
gcc version 4.0.0 20050406 (Red Hat 4.0.0-0.41.fc3)

g++4  with -O2 blows on some files with Internal Compiler Error
(see http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20991)
so I had to recompile some files manually with -Os.
The end result looks OK. Passes all tests etc....

I guess my point is that gcc4 pre-release may be finicky with
optimization.

> Thanks,
>
> jwe

Sincerely,

Dmitri.
--



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------