graphviz, graphviz-devel: configure script doesn't like -headerpad_max_install_names

[cooljeanius (Eric Gallager)]

As of MacPorts 2.2.0, -Wl,-headerpad_max_install_names is now added to the default configure.ldflags. However, Graphviz's configure script complains about this:

checking if intptr_t is declared... yes
checking if assert needs __eprintf... i686-apple-darwin10-gcc-4.2.1: -headerpad_max_install_names: linker input file unused because linking not done
i686-apple-darwin10-gcc-4.2.1: -headerpad_max_install_names: linker input file unused because linking not done
no

This is just a warning, and configuring continues, but it still might be worth considering whether to keep the default configure.ldflags for Graphviz.

[ryandesign (Ryan Carsten Schmidt)]

Yes, I want graphviz (and as many other ports as possible) to build relocatable libraries. That is why I added those flags to the default configure.ldflags. If the port does in fact build relocatably using these flags, and this is only a configure warning, then I'm not concerned in the slightest, though I will report the problem to the developers of graphviz.

[cooljeanius (Eric Gallager)]

I think this is the offending part of the configure.ac file:

AC_MSG_CHECKING(if assert needs __eprintf)
cat > conftest.$ac_ext <<EOF
#include "confdefs.h"
#include "assert.h"
int main(int argc, char **argv) {
assert(argc);
; return 0; }
EOF
${CC-cc} -c $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext 1>&5
if test "x`nm conftest.o |$EGREP __eprintf`" != "x"; then
    AC_MSG_RESULT(yes)
    CFLAGS="$CFLAGS -DNDEBUG"
else
    AC_MSG_RESULT(no)
fi
rm -f conftest*

I bet the graphviz developers could probably think of a better way to do this...

[ryandesign (Ryan Carsten Schmidt)]

Replying to ryandesign@…:

I will report the problem to the developers of graphviz.

​http://www.graphviz.org/mantisbt/view.php?id=2325

[ryandesign (Ryan Carsten Schmidt)]

Fixed in graphviz 2.34.0.