> According to Matěj Týč on 3/23/2009 5:25 PM:
> > Hello, I have found out that having this in configure.ac is a problem:
> >
> > AC_DEFUN([foo],
> > [echo "#" ])
>
> The bug is in your code - # starts a comment in m4, hence, the ]) is still
> part of the comment, and you haven't finished th
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Matěj Týč on 3/24/2009 5:20 AM:
> Regarding the manual: I have If you look at the page 158 (AS_TR_CPP
> description), there is an unquoted hash (#) used in a configure.ac-like
> file without any warning.
Thanks for catching that; it is i
Hello,
I have a Fortran program that uses LAPACK. Since LAPACK can come from
different vendors depending on the platform,
and be in different locations (e.g., /usr/lib/liblapack.a with the reference
implementation, or in mkl_lapack.a in
the Intel compiler tree when using the Intel compiler),
Hello,
I think I solved one of my problems. Evidently, there's an Autoconf
macro AC_LANG. By adding AC_LANG(Fortran) to my configure.ac file, it
ensures that test programs are written in Fortran rather than in C, and
that seems sufficient to find sgetrf in liblapack.
Now, onto the next issue!
On 24-Mar-2009, David A. Ventimiglia wrote:
| I think I solved one of my problems. Evidently, there's an Autoconf
| macro AC_LANG. By adding AC_LANG(Fortran) to my configure.ac file, it
| ensures that test programs are written in Fortran rather than in C, and
| that seems sufficient to find sget
When trying to build sane-backends in Mac OS X on an Intel iMac, I ran
into a problem (see below). It appears that the configure script does
not cope with spaces in the name of a directory in the path to the
directory in which it resides.
The opinion on the sane-devel mailing list was that
