[physfs] Fix PhysicsFS 1.0.2 build error on Mac OS X

Yuji Yamano yyamano at kt.rim.or.jp
Mon Aug 24 23:40:48 EDT 2009


On Mon, 24 Aug 2009 08:46:11 -0400, "Ryan C. Gordon" <icculus at icculus.org> wrote:

> We're probably not going to do another 1.0.x release, 

It's ok for me. I commited the diff as pkgsrc patch.

http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/devel/physfs/patches/patch-ab?rev=1.5&content-type=text/x-cvsweb-markup&only_with_tag=MAIN

> so it's not worth fixing this properly (2.0 doesn't use configure.in
> anymore), but this patch would make PhysicsFS itself depend on
> readline, when it's just an optional feature of the test program.

I think the dependency is added only if readline is enabled.

if test x$enable_readline = xyes; then
  AC_CHECK_HEADER(readline/readline.h, have_readline_hdr=yes)
  AC_CHECK_LIB(readline, readline, have_readline_lib=yes, , -lcurses)
  AC_CHECK_HEADER(readline/history.h, have_history_hdr=yes)
  AC_CHECK_LIB(readline, add_history, have_history_lib=yes, , -lcurses)
  if test x$have_readline_hdr = xyes -a x$have_readline_lib = xyes; then
    if test x$have_history_hdr = xyes -a x$have_history_lib = xyes; then
      AC_DEFINE([PHYSFS_HAVE_READLINE], 1, [define if we have readline])
      have_readline="yes"
      LIBS="-lreadline -lcurses $LIBS"
    fi
  fi
fi

> I'm not sure why Mac OS X is ignoring the BUILD_READLINE setting, but
> it's probably another esoteric automake quirk. It's why we moved to
> CMake.

I'm not sure if it's a Mac OS X specific issue. I can't find where
"-lreadline -lcurses" is added to LIBS in configure, but I'm not an
automake/configure expert.

-- Yuji Yamano
Peace, unity, love and having fun!




More information about the physfs mailing list