hi ben, On Mon, Mar 02, 2009 at 09:14:36PM -0800, Ben Pfaff wrote: > > # SM needs ICE to (dynamically) link under SunOS 4.x (so we have to > > # check for ICE first), but we must link in the order -lSM -lICE or > > # we get undefined symbols. So assume we have SM if we have ICE. > > # These have to be linked with before -lX11, unlike the other > > # libraries we check for below, so use a different variable. > > # John Interrante, Karl Berry > > AC_CHECK_LIB(ICE, IceConnectionNumber, > > [X_PRE_LIBS="$X_PRE_LIBS -lSM -lICE"], , $X_EXTRA_LIBS) > > LDFLAGS=$ac_save_LDFLAGS > > > > it seems there ought to be a better test for that...
> In other words, I would like to do something about this (or to
> ask the upstream Autoconf developers to do so) but I don't see a
> way to do it without breaking other things. Suggestions are
> welcome, of course...
yeah. my first inclination was to suggest gutting it entirely,
but then i realized it's likely someone would want to generate a
./configure file from a debian system to be used on one of these old
(sunos 4.x???) systems.
as i understand it, this chunk of code is because linking -lICE requires
first linking -lSM on some platforms, but it unconditionally assumes the
need to link against -lICE. so therefore wouldn't it be possible to
check if it's needed in the first place (i.e. if there's no -lICE in the
current lib flags)?
otherwise since sunos 4.x is the only mentioned system to have this
problem, one could perhaps put a wrapper around the current code testing
for that...? and yeah, perhaps upstream would be better to make the
call on this.
sean
signature.asc
Description: Digital signature
