Re: Autoconf fooled by AC_FC_LINE_LENGTH

[Peter Rosin <peda at lysator dot liu dot se>]

On 2016-07-29 14:23, Marco Atzeri wrote:
> Hi Eric,
> 
> building latest Arpack
> https://github.com/opencollab/arpack-ng/releases
> https://github.com/opencollab/arpack-ng/issues/41
> 
> I hit a strange case where the presence of a single line
> 
> AC_FC_LINE_LENGTH(unlimited)
> 
> before the usual
> LT_PREREQ([2.4.2])
> LT_INIT([win32-dll])
> 
> is causing the libtool to be incorrectly set, versus
> the right case
> ---------------------------------------------------
> -libext=a
> +libext=lib
> 
> -shlibpath_overrides_runpath=yes
> +shlibpath_overrides_runpath=unknown
> 
> -library_names_spec="\$libname.dll.a"
> +library_names_spec="\$libname`echo \$release | \$SED -e 
> s/[.]/-/g`\$versuffix\$shared_ext \$libname.lib"
> 
> -soname_spec="`echo \$libname | sed -e s/^lib/cyg/``echo \$release | 
> \$SED -e s/[.]/-/g`\$versuffix\$shared_ext"
> +soname_spec=""
> -----------------------------------------------------
> 
> It seems libtool is incorrectly set for MSVC platform
> 
> I tried to move the libtool inizialization before the
> AC_FC_LINE_LENGTH but autoconf is objecting on wrong preference.
> 
> It seems a bug impacting only cygwin.

Reading the source comment for AC_FC_LINE_LENGTH indicates that
you should call AC_FC_FREEFORM or AC_FC_FIXEDFORM before you call
AC_FC_LINE_LENGTH. You do not do that. This is not mentioned in
the autoconf manual I'm looking at. I don't know if the source
code comment or the manual is correct (it's been 20 years since I
last used fortran...)

However, I suspect the issue is that libtool might eat or be
otherwise confused by the flag selected to get fortran support
for long lines. What is FCFLAGS when you have the macro present?

Out on a limb here...

Cheers,
Peter

PS. some logs would be handy...

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple