===================================================================
RCS file: /cvsrepo/anoncvs/cvs/src/usr.bin/lorder/lorder.sh,v
retrieving revision 1.9
retrieving revision 1.10
diff -u -r1.9 -r1.10
--- src/usr.bin/lorder/lorder.sh	1998/11/22 00:11:08	1.9
+++ src/usr.bin/lorder/lorder.sh	1999/05/21 01:24:04	1.10
@@ -1,6 +1,6 @@
 #!/bin/sh -
-#	$OpenBSD: lorder.sh,v 1.9 1998/11/22 00:11:08 d Exp $
-#	$NetBSD: lorder.sh,v 1.3 1995/04/24 07:38:52 cgd Exp $
+#	$OpenBSD: lorder.sh,v 1.10 1999/05/21 01:24:04 espie Exp $
+#	$NetBSD: lorder.sh.gnm,v 1.3 1995/12/20 04:45:11 cgd Exp $
 #
 # Copyright (c) 1990, 1993
 #	The Regents of the University of California.  All rights reserved.
@@ -36,7 +36,7 @@
 #	@(#)lorder.sh	8.1 (Berkeley) 6/6/93
 #
 
-# only one argument is a special case, just output the name twice
+# one argument can be optimized: put out the filename twice
 case $# in
 	0)
 		echo "usage: lorder file ...";
@@ -62,23 +62,16 @@
 # remove temporary files on HUP, INT, QUIT, PIPE, TERM
 trap "rm -rf $TDIR; trap 2 ; kill -2 $$" 1 2 3 13 15
 
-# if the line ends in a colon, assume it's the first occurrence of a new
-# object file.  Echo it twice, just to make sure it gets into the output.
+# make sure files depend on themselves
+for file in "$@"; do echo "$file $file" ; done
+# if the line has " T ", " D ", " G ", " R ",  it's a globally defined 
+# symbol, put it into the symbol file.
 #
-# if the line has " T " or " D " it's a globally defined symbol, put it
-# into the symbol file.
-#
 # if the line has " U " it's a globally undefined symbol, put it into
 # the reference file.
 ${NM:-nm} -go "$@" | sed "
-	/:$/ {
-		s/://
-		s/.*/& &/
-		p
-		d
-	}
-	/ [TD] / {
-		s/:.* [TD] / /
+	/ [TDGR] / {
+		s/:.* [TDGR] / /
 		w $S
 		d
 	}
@@ -90,7 +83,7 @@
 "
 
 # sort symbols and references on the first field (the symbol)
-# join on that field, and print out the file names.
+# join on that field, and print out the file names (dependencies).
 sort +1 $R -o $R
 sort +1 $S -o $S
 join -j 2 -o 1.1 2.1 $R $S