octave-maintainers
[Top][All Lists]
Advanced

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

Re: problem building octave 3.0 on x86_64


From: John W. Eaton
Subject: Re: problem building octave 3.0 on x86_64
Date: Fri, 28 Mar 2008 00:09:53 -0400

On 27-Mar-2008, Dmitri A. Sergatskov wrote:

| On Thu, Mar 27, 2008 at 9:58 PM, gbohannon <address@hidden> wrote:
| >
| >  I had the same problem in RHEL5 today. Please let me know if the problem is
| >  solved because I'd love to use the new Octave. I am running the 64-bit WS
| >  version of RHEL5, and I think I have the full development suite of software
| >  provided by Red Hat.
| 
| >  -L/usr/lib/../lib64 -lz -lfrtbegin -lg2c -lm
| >  ../src/liboctinterp.so: undefined reference to `__cxa_get_exception_ptr'
| 
| I am shooting in the dark, but I suspect that you used g77 as your
| fortran compiler.
| Do you have gfortran available on RHEL5?
| If so try:
| 
| ./configure --with-f77=gfortran
| 
| or
| 
| FC=gfortran ./configure
| 
| and see if that works better...

Oh, now I seem to remember that something like this was reported once
before.

Now that gfortran is the default compiler in the compiler collection,
the autoconf Fortran compiler detection macro should probably have the
list of compilers modified to put gfortran in front of g77, or at
least to do that if the selected version of gcc/g++ is newer than 4.?
(I'm not sure what ? should be).  Perhaps someone can report this
problem to autoconf maintainers, though I'm not sure they will do
anything since autoconf treats gfortran as an f95 compiler and g77 as
an f77 compiler, and since we are using the f77 check, the autoconf
macro will look for f77 compilers before trying f90 or f95 compilers.
Should we be checking for f95 compilers instead?

jwe



reply via email to

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