Re: fail.at woes

[Bob Friesenhahn]

On Fri, 3 Feb 2006, Ralf Wildenhues wrote:

> fail.at fails on IRIX and AIX: their linkers will warn about duplicate
> symbols when creating a shared library, but they will then go on to
> happily discard the duplicate definitions and exit with zero.
>
> So, since the aim of the test is not to exercise the linker, but the
> libtool script failure mode, we'll just also pass a bogus object file
> to the linker, to make it barf more reliably.  ;-)
>
> OK to apply?

Please do!

Bob

>        * tests/fail.at: Also pass a bogus object file to the linker so
>        it fails more reliably; AIX and IRIX linkers ignore duplicate
>        symbol definitions.
>
> Index: tests/fail.at
> ===================================================================
> RCS file: /cvsroot/libtool/libtool/tests/fail.at,v
> retrieving revision 1.1
> diff -u -r1.1 fail.at
> --- tests/fail.at       3 Dec 2005 08:33:48 -0000       1.1
> +++ tests/fail.at       3 Feb 2006 09:05:13 -0000
> @@ -61,7 +62,16 @@
>   echo 'double duplicate_name(double x) { return 2.*x; }' > b.c
>   $LIBTOOL --mode=compile $CC $CPPFLAGS $CFLAGS -c a.c
>   $LIBTOOL --mode=compile $CC $CPPFLAGS $CFLAGS -c b.c
> -  FAIL_CHECK([$LIBTOOL --mode=link $CC $CFLAGS $LDFLAGS -o liba.la a.lo b.lo 
> -rpath /foo])
> +  # Unfortunately, this may actually succeed on AIX and IRIX  :(
> +  # So, also add a bogus object file to provoke more failure.
> +  echo 'int whatever(void) { return 0;}' > c.c
> +  $LIBTOOL --mode=compile $CC $CPPFLAGS $CFLAGS -c c.c
> +  (
> +    . ./c.lo
> +    test "$pic_object" != none && echo choke me >"$pic_object"
> +    test "$non_pic_object" != none && echo choke me >"$non_pic_object"
> +  )
> +  FAIL_CHECK([$LIBTOOL --mode=link $CC $CFLAGS $LDFLAGS -o liba.la a.lo b.lo 
> c.lo -rpath /foo])
>   AT_CHECK([test -f liba.la], [1])
>   ;;
> esac

======================================
Bob Friesenhahn
address@hidden, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer,    http://www.GraphicsMagick.org/