>Number:         30051
>Category:       bin
>Synopsis:       iconv(1) error message is misleading
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    bin-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Apr 24 17:47:00 +0000 2005
>Originator:     jklowden@schemamania.org
>Release:        NetBSD 2.0
>Organization:

>Environment:
System: NetBSD oak.schemamania.org 2.0 NetBSD 2.0 (GENERIC) #0: Wed Dec 1 10:58:25 UTC 2004 builds@build:/big/builds/ab/netbsd-2-0-RELEASE/i386/200411300000Z-obj/big/builds/ab/netbsd-2-0-RELEASE/src/sys/arch/i386/compile/GENERIC i386
Architecture: i386
Machine: i386

$ ident `which iconv`
/usr/bin/iconv: # sorry, the binary seems not to have an ident string? 
     $NetBSD: crt0.c,v 1.13 2003/07/26 19:24:27 salo Exp $

>Description:
Given bad input, __iconv() returns EINVAL, which is good.   But
when iconv(1) calls errx with EINVAL, the user sees:

	iconv: iconv(): Invalid argument
	 
which to my mind a reasonable user would interpret as meaning the command-line
options were wrong, not that the inputs were bad.  If it called errx with
EILSEQ, you'd see "iconv: iconv(): Illegal byte sequence", which at least points
the user at the data.  

>How-To-Repeat:
	$ echo 'hi' | iconv -t ascii -f ucs-2le
	iconv: iconv(): Invalid argument
>Fix:
Well, here's my suggestion.  
Index: iconv.c
===================================================================
RCS file: /cvsroot/src/usr.bin/iconv/iconv.c,v
retrieving revision 1.6
diff -u -r1.6 iconv.c
--- iconv.c	21 Dec 2004 11:33:07 -0000	1.6
+++ iconv.c	24 Apr 2005 12:03:34 -0000
@@ -124,9 +128,9 @@
 					    INBUFSIZE-inbytes, fp);
 				if (ret == 0) {
 					if (feof(fp))
-						errx(EXIT_FAILURE,
-						     "iconv(): %s",
-						     strerror(EINVAL));
+						errx(EXIT_FAILURE, 
+						     "iconv(): %s", 
+						     strerror(EILSEQ));
 					else
 						err(EXIT_FAILURE, "fread()");
 				}