[PATCH]: A new one for ld/Makefile.am

["H . J . Lu" <hjl at valinux dot com>]

On Fri, Mar 10, 2000 at 12:07:42AM -0300, Alexandre Oliva wrote:
> On Mar  9, 2000, "H . J . Lu" <hjl@valinux.com> wrote:
> 
> > On Thu, Mar 09, 2000 at 11:12:42PM -0300, Alexandre Oliva wrote:
> >> On Mar  9, 2000, "H . J . Lu" <hjl@valinux.com> wrote:
> >> 
> >> >> all: stmp-run-ld
> >> >> stmp-run-ld: ld-new
> >> >> @ ./ld-new > $@ 2>&1
> >> 
> >> > Tried. It doesn't work with parallel build due to the recursive make.
> >> 
> >> Use `all-am', then.
> 
> > It works a little better. But it still doesn't work right since
> > ld-new$(EXEEXT) is made through the recursive make.
> 
> I can't see any recursive makes involving ld-new other than the one
> automake itself creates, and the target of that recursive make is
> all-am.  Am I missing anything?
> 

It was my mistake. It works. Here is a new patch for ld/Makefile.am.
Any objections?

Thanks.


H.J.
----
2000-03-09  H.J. Lu  <hjl@gnu.org>

	* Makefile.am (stmp-run-ld): New target.
	(all-am): Depend on it.
	(clean-stmp-run-ld): New target.
	(clean): Depend on it.
	* Makefile.in: Rebuild.

--- Makefile.am.orig	Thu Mar  9 10:23:09 2000
+++ Makefile.am	Thu Mar  9 20:43:30 2000
@@ -674,6 +674,20 @@ check-DEJAGNU: site.exp
 	else echo "WARNING: could not find \`runtest'" 1>&2; :;\
 	fi
 
+# We want to run the new linker once to create .libs/lt-ld-new if
+# necessary. Otherwise, the new linker may not work. We don't care
+# if it really works or not.
+stmp-run-ld: ld-new$(EXEEXT)
+	-./ld-new$(EXEEXT) -v > $@ 2>&1
+	touch $@
+
+all-am: stmp-run-ld
+
+clean-stmp-run-ld:
+	-rm -f stmp-run-ld
+
+clean-am: clean-stmp-run-ld
+
 # Rules for testing by relinking ld itself.
 # A similar test is in the testsuite.  This target is for ease of use
 # when porting ld.