diff --git a/GNUmakefile.in b/GNUmakefile.in
index f3e33fb2788d48b509b8df9e5b9f9c2d12b6d465..7c387333260c83ea2d36f1deeb69107857cbf13e 100644
--- a/GNUmakefile.in
+++ b/GNUmakefile.in
@@ -1,7 +1,7 @@
 #
 # PostgreSQL top level makefile
 #
-# $Header: /cvsroot/pgsql/GNUmakefile.in,v 1.3 2000/06/10 18:01:34 petere Exp $
+# $Header: /cvsroot/pgsql/GNUmakefile.in,v 1.4 2000/06/11 18:43:52 tgl Exp $
 #
 
 srcdir = @srcdir@
@@ -36,11 +36,20 @@ ACLOCAL = @ACLOCAL@
 GNUmakefile: GNUmakefile.in $(top_builddir)/config.status
 	CONFIG_FILES=$@ CONFIG_HEADERS= ./config.status
 
-$(top_builddir)/config.status: $(top_srcdir)/configure
-	cd $(top_builddir) && ./config.status --recheck
+# This rule does not work --- what if config.status doesn't exist?
 
-$(top_srcdir)/configure: $(top_srcdir)/configure.in $(top_srcdir)/aclocal.m4
-	cd $(top_srcdir) && $(AUTOCONF)
+# $(top_builddir)/config.status: $(top_srcdir)/configure
+# 	cd $(top_builddir) && ./config.status --recheck
 
-$(top_srcdir)/aclocal.m4: $(wildcard $(top_srcdir)/config/*.m4)
-	cd $(top_srcdir) && $(ACLOCAL) -I config
+# These dependencies are evil and dangerous, because they can cause make
+# to re-run autoconf and then re-run configure due to configure not
+# having a newer timestamp than configure.in after a CVS pull.  Same
+# problem for aclocal timestamp skew.  This solution is considerably
+# worse than the problem it was intended to solve.
+# Do not put it back or I will take it right out again --- tgl
+
+# $(top_srcdir)/configure: $(top_srcdir)/configure.in $(top_srcdir)/aclocal.m4
+# 	cd $(top_srcdir) && $(AUTOCONF)
+
+# $(top_srcdir)/aclocal.m4: $(wildcard $(top_srcdir)/config/*.m4)
+# 	cd $(top_srcdir) && $(ACLOCAL) -I config