From owner-freebsd-bugs  Tue Jun  4 11:41:03 1996
Return-Path: owner-bugs
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id LAA02896
          for bugs-outgoing; Tue, 4 Jun 1996 11:41:03 -0700 (PDT)
Received: from mail.cs.tu-berlin.de (root@mail.cs.tu-berlin.de [130.149.17.13])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id LAA02780
          for <bugs@freebsd.org>; Tue, 4 Jun 1996 11:39:08 -0700 (PDT)
Received: from campa.panke.de (anonymous224.ppp.cs.tu-berlin.de [130.149.17.224]) by mail.cs.tu-berlin.de (8.6.12/8.6.12) with ESMTP id UAA03497; Tue, 4 Jun 1996 20:13:17 +0200
Received: (from wosch@localhost) by campa.panke.de (8.6.12/8.6.12) id TAA01503; Tue, 4 Jun 1996 19:27:07 +0200
Date: Tue, 4 Jun 1996 19:27:07 +0200
From: Wolfram Schneider <wosch@cs.tu-berlin.de>
Message-Id: <199606041727.TAA01503@campa.panke.de>
To: joerg_wunsch@uriah.heep.sax.de (Joerg Wunsch)
Cc: jin@george.lbl.gov (Jin Guojun[ITG]), bugs@freebsd.org, problem@bsdi.com
Subject: Re: sh bug
In-Reply-To: <199606040947.LAA21894@uriah.heep.sax.de>
References: <199606031939.MAA26616@george.lbl.gov>
	<199606040947.LAA21894@uriah.heep.sax.de>
Reply-to: Wolfram Schneider <wosch@cs.tu-berlin.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
Sender: owner-bugs@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

J. Wunsch writes:
>Our test(1) implementation fully agrees with Posix.2:

agrees != do the right thing

BTW,
$ /bin/test \( e = -e \)
test: syntax error: Undefined error: 0


I think FreeBSD should change the implementation of test(1).

Wolfram

Index: test.c
===================================================================
RCS file: /a/cvs/src/bin/test/test.c,v
retrieving revision 1.13
diff -u -r1.13 test.c
--- 1.13	1996/03/11 11:01:03
+++ test.c	1996/06/04 16:34:24
@@ -198,7 +198,9 @@
 		if (opname[0] == '(' && opname[1] == '\0') {
 			nest += NESTINCR;
 			continue;
-		} else if (*ap && (op = lookup_op(opname, unary_op)) >= 0) {
+		} else if (*ap && (op = lookup_op(opname, unary_op)) >= 0 &&
+			   lookup_op(*ap, binary_op) < 0 && 
+			   strcmp(*ap, ")" )) {
 			if (opsp == &opstack[0])
 				overflow();
 			--opsp;




>
> The algorithm for determining the precedence of the operators and the      1
> return value that shall be generated is based on the number of arguments   1
> presented to test.  (However, when using the [...] form, the right-        1
> bracket final argument shall not be counted in this algorithm.)  In the    1
> following list, $1, $2, $3, and $4 represent the arguments presented to    1
> test.                                                                      1
>
>     0 arguments:                                                           1
>...
>     1 argument:                                                            1
>...
>     2 arguments:                                                           1
>...
>     3 arguments:                                                           1
>
>           - If $2 is a binary primary, perform the binary test of $1 and   2
>             $3.                                                            2
>
>           - If $1 is !, negate the two-argument test of $2 and $3.         1
>
>           - Otherwise, produce unspecified results.                        1
>
>     4 arguments:                                                           1
>
>           - If $1 is !, negate the three-argument test of $2, $3, and $4.  1
>
>           - Otherwise, the results are unspecified.                        1
>
>    >4 arguments:                                                           1
>
>          The results are unspecified.                                      1
>
>
>The ``> 4 arguments'' is the case you are complaining about.  As you
>can see, it's up to the implementation of the test command to decide
>about the operator precedence.  It apparently decides that the first
>operator is to be taken as the -h flag (test for a symbolic link), and
>the following equation sign is taken as the filename argument to -h.
>The remaining arguments finally result in a syntax error.