Bug 13363

Unix 98 requires that env(1) take the option -i to destroy the
inherited environment.  This would be identical to the currently
existing - option.  This diff adds a fallthrough so that -i is
handled as Unix 98 specifies.  This does not include a change for
the man page, I did not think it was necessary.

How-To-Repeat: 
env -i gives error about illegal option

On Tue, 24 Aug 1999, Bill Fumerola wrote:

> style(9) says this can be written as
> 
> 			/* For Unix 98 compatability. */
> 			/* FALLTHROUGH */

Okay, here is a modified diff.  I also mispelled "compatibility" so I
fixed that too.

*** /usr/src/usr.bin/env/env.c	Sun Dec  6 22:58:22 1998
--- /usr/local/src/env/env.c	Tue Aug 24 20:33:03 1999
***************
*** 60,67 ****
  	char *cleanenv[1];
  	int ch;
  
! 	while ((ch = getopt(argc, argv, "-")) != -1)
  		switch(ch) {
  		case '-':
  			environ = cleanenv;
  			cleanenv[0] = NULL;
--- 60,72 ----
  	char *cleanenv[1];
  	int ch;
  
! 	while ((ch = getopt(argc, argv, "i-")) != -1)		
  		switch(ch) {
+ 		case 'i':
+ 			/* For Unix 98 compatibility. */
+ 			/* FALLTHROUGH */
  		case '-':
  			environ = cleanenv;
  			cleanenv[0] = NULL;

On Tue, 24 Aug 1999 22:42:09 -0400, James Howard wrote:

> ! 	while ((ch = getopt(argc, argv, "i-")) != -1)		
>   		switch(ch) {
> + 		case 'i':
> + 			/* 
> + 			 * For Unix 98 compatability.
> + 			 */
> + 			/* FALLTHROUGH */

Hi James/Jamie (which do you prefer?),

Bill Fumerola replied telling you that style(9) requires a slightly
different format for your comments. While it's true, the example he gave
you doesn't appear in style(9).

I'd do this:

		case 'i':		/* Unix 98 compatibility option */
			/* FALLTHROUGH */

Personally, I wouldn't provide the comment at all. Focus on the manpage
instead. :-P

Oh, and note the spelling of compatibility. :-)

later,
Sheldon.

Due to the sheer number of complaints about my style here is a new patch
for env which should make everyone happy :)  Following it is a diff to
printenv.1.

--- /usr/src/usr.bin/env/env.c	Sun Dec  6 17:58:22 1998
+++ env.c	Wed Aug 25 17:45:39 1999
@@ -60,8 +60,9 @@
 	char *cleanenv[1];
 	int ch;
 
-	while ((ch = getopt(argc, argv, "-")) != -1)
+	while ((ch = getopt(argc, argv, "i-")) != -1)		
 		switch(ch) {
+		case 'i':
 		case '-':
 			environ = cleanenv;
 			cleanenv[0] = NULL;
@@ -85,6 +86,6 @@
 usage()
 {
 	(void)fprintf(stderr,
-	    "usage: env [-] [name=value ...] [command]\n");
+	    "usage: env [-] [-i] [name=value ...] [command]\n");
 	exit(1);
 }

------------CUT HERE-----------------

--- /usr/src/usr.bin/printenv/printenv.1	Mon Jul 12 16:23:52 1999
+++ printenv.1	Wed Aug 25 17:45:42 1999
@@ -45,6 +44,7 @@
 .Op Ar name
 .Nm env
 .Op Fl
+.Op Fl i
 .Op Ar name=value ...
 .Op Ar command
 .Sh DESCRIPTION
@@ -72,12 +72,20 @@
 .Ar name  ,
 with a value of
 .Ar value  .
-The option
-.Sq Fl
-causes
+The options to
+.Nm env
+are as follows:
+.Pp
+.Bl -tag -width indent
+.It Fl
+Force
 .Nm env
 to completely ignore the environment
 it inherits.
+.It Fl i
+Equivalent to
+.Fl .
+.El
 .Pp
 If no command is specified,
 .Nm env

On Wed, 25 Aug 1999 17:57:56 -0400, James Howard wrote:

> -	while ((ch = getopt(argc, argv, "-")) != -1)
> +	while ((ch = getopt(argc, argv, "i-")) != -1)		
>  		switch(ch) {
> +		case 'i':
>  		case '-':

You don't seem to have addressed the case disordering that Bruce
commented on. :-)

> +.It Fl i
> +Equivalent to
> +.Fl .

This text begs the question "why?". I'd also suggest that the behaviour
of the two options be described under -i, with - being declared the
compatibility option. This should encourage people to use -i for
portability. This is how the Single UNIX Spec does it. :-P

Ciao,
Sheldon.

Index: usr.bin/printenv/printenv.1
===================================================================
RCS file: /home/ncvs/src/usr.bin/printenv/printenv.1,v
retrieving revision 1.3
diff -u -d -r1.3 printenv.1
--- printenv.1	1999/07/12 20:23:52	1.3
+++ printenv.1	1999/08/26 13:02:02
@@ -45,6 +45,7 @@
 .Op Ar name
 .Nm env
 .Op Fl
+.Op Fl i
 .Op Ar name=value ...
 .Op Ar command
 .Sh DESCRIPTION
@@ -72,12 +73,21 @@
 .Ar name  ,
 with a value of
 .Ar value  .
-The option
-.Sq Fl
-causes
+.Pp
+The options are as follows:
+.Bl -tag -width indent
+.It Fl i
+Execute the
+.Ar command
+with only those environment values specified. The environment inherited
+by
 .Nm env
-to completely ignore the environment
-it inherits.
+is ignored completely.
+.It Fl
+Identical to the
+.Fl i
+option, available for backward compatibility.
+.El
 .Pp
 If no command is specified,
 .Nm env
Index: usr.bin/env/env.c
===================================================================
RCS file: /home/ncvs/src/usr.bin/env/env.c,v
retrieving revision 1.4
diff -u -d -r1.4 env.c
--- env.c	1998/12/06 22:58:22	1.4
+++ env.c	1999/08/26 12:57:34
@@ -60,9 +60,10 @@
 	char *cleanenv[1];
 	int ch;
 
-	while ((ch = getopt(argc, argv, "-")) != -1)
+	while ((ch = getopt(argc, argv, "-i")) != -1)
 		switch(ch) {
 		case '-':
+		case 'i':
 			environ = cleanenv;
 			cleanenv[0] = NULL;
 			break;
@@ -85,6 +86,6 @@
 usage()
 {
 	(void)fprintf(stderr,
-	    "usage: env [-] [name=value ...] [command]\n");
+	    "usage: env [-] [-i] [name=value ...] [command]\n");
 	exit(1);
 }

Comment 6 Sheldon Hearn 1999-08-26 14:11:47 UTC

Responsible Changed
From-To: freebsd-bugs->sheldonh

I'll take this one. 

On Thu, 26 Aug 1999, Sheldon Hearn wrote:

> This text begs the question "why?". I'd also suggest that the behaviour
> of the two options be described under -i, with - being declared the
> compatibility option. This should encourage people to use -i for
> portability. This is how the Single UNIX Spec does it. :-P

I completely agree with this but I didn't want to push it down their
throats.  As I said previously, the draft of XPG6 removes ``-'' in favor
of ``-i''.  To me, this means don't even bother documenting ``-'' and
leave it in quietly as a compatibility option.  I personally view it as a
bad idea to to have done ``-'' in the first place.  However, I was 7 when
this was first done (possibly younger).  I will not question it further :)

Jamie

Comment 8 Sheldon Hearn 1999-08-27 09:59:38 UTC

State Changed
From-To: open->analyzed

Committed to CURRENT. I'll leave it in this state for MFC. 

Comment 9 Sheldon Hearn 1999-12-20 16:38:29 UTC

State Changed
From-To: analyzed->closed

Damn, I left this in the wrong state for MFC.  Since it missed 
3.4-RELEASE, and since there isn't going to be another release 
thereafter on the RELENG_3 line, an MFC now  would be silly.