(change has no real effect, but I found the inconsistency when lookin…

…g for

a real problem which wasn't here...)
	* write_msg.c (krb5_write_message): check the real length
	variable, not the net version.


git-svn-id: svn://anonsvn.mit.edu/krb5/trunk@6842 dc483132-0cff-0310-8789-dd5450dbe970

src/lib/krb5/os/ChangeLog

@@ -1,3 +1,8 @@
+Fri Sep 22 15:57:33 1995  Mark Eichin  <eichin@cygnus.com>
+
+	* write_msg.c (krb5_write_message): check the real length
+	variable, not the net version.
+
 Wed Sep 13 11:01:18 1995 Keith Vetter (keithv@fusion.com)
 
 	* Makefile.in: had to define away DBFLAGS for the PC.
@@ -131,7 +136,6 @@ Fri Jun 30 14:44:17 EDT 1995	Paul Park	(pjpark@mit.edu)
 		DBM aname database format.
 	* Makefile.in - Add definitions to select aname rules.
 
-
 Tue Jun 27 15:48:41 EDT 1995	Paul Park	(pjpark@mit.edu)
 	* read_pwd.c(intr_routine) - Add signal name argument to signal handler
 		to conform to prototype.

src/lib/krb5/os/write_msg.c

@@ -41,7 +41,7 @@ krb5_write_message(context, fdp, outbuf)
 	if (krb5_net_write(context, fd, (char *)&len, 4) < 0) {
 		return(errno);
 	}
-	if (len && (krb5_net_write(context, fd, outbuf->data, outbuf->length) < 0)) {
+	if (outbuf->length && (krb5_net_write(context, fd, outbuf->data, outbuf->length) < 0)) {
 		return(errno);
 	}
 	return(0);