Re: libtool-branch-2-0: add strlcat/strlcpy emulation

[Ralf Wildenhues]

* Bob Friesenhahn wrote on Mon, Dec 13, 2004 at 05:00:13PM CET:
> On Mon, 13 Dec 2004, Ralf Wildenhues wrote:
> >>
> >>AC_REPLACE_FUNCS doesn't create an empty .o file like yours does. But,
> >>this is I think only a problem on Tru64 UNIX which deosn't have
> >>strlcat.
> >
> >Is there a reason against using
> >
> >#if defined(HAVE_STRLCAT) && defined(HAVE_STRLCPY)
> >static int dummy;
> >#endif
> >
> >in libltdl/lt__strl.c besides the apparent uglyness?  (Did I mention
> >before that I think that the LIBOBJ mechanism less than optimal?)
> 
> This might also result in an empty object file, or a warning that a 
> variable was not used.

Yep, depending on the system.  Since I now decided to check rather than
guess (sorry for not doing that earlier):  We are using something like
this already for lt__dirent in m4/ltdl.m4.  Could you just add similarly
for lt__strl?

Sorry for so much noise, and apologies to you and Albert for insisting
on a stupid idea.

> Is an empty object file a problem for some compilers/linkers?  Do they 
> complain or refuse to work?

Yes, they fail (AIX ld does for example).

Regards,
Ralf