Re: sh compatibility issue

[Peter Stephenson <p.w.stephenson@ntlworld.com>]

On Tue, 22 Feb 2011 17:51:18 -0800
Bart Schaefer <schaefer@brasslantern.com> wrote:
> On Tuesday, February 22, 2011, Vincent Stemen <vince.lists@hightek.org> wrote:
> >
> > I don't think it worked.  With this patch added, the bahaviour did not
> > change from the previous patch.  Fatal error in sh mode not not in zsh
> > mode.
> 
> That's intentional.  Zsh is not a POSIX shell unless you tell it to
> be, so in zsh mode it behaves the way zsh always has.
> 
> As I understand it, the patch should change the behavior of other
> special builtins (in addition to exec), but only when POSIXBUILTINS is
> set.

Bart's right, sorry, I wasn't clear enough.  Yes, this is the
long-standing policy.

If we'd been designing zsh from scratch now, there would be no point in
all these small areas where it's a bit different from POSIX by accident
rather than design.  As it is, given the shell's manifold differences in
native mode, there's no point squeezing extra POSIX-compliance out of
native mode at the expense of backward compatibility.

By the way, one small tweak which should be rarely visible: now I've
upped the ante, there may be cases where we're forked at the point we
need to exit.

Index: Src/exec.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/exec.c,v
retrieving revision 1.190
diff -p -u -r1.190 exec.c
--- Src/exec.c	22 Feb 2011 20:09:20 -0000	1.190
+++ Src/exec.c	23 Feb 2011 09:21:32 -0000
@@ -3306,8 +3306,10 @@ execcmd(Estate state, int input, int out
 	 */
 	if (redir_err || errflag) {
 	    if (!isset(INTERACTIVE)) {
-		/* We've already _exit'ed if forked */
-		exit(1);
+		if (forked)
+		    _exit(1);
+		else
+		    exit(1);
 	    }
 	    errflag = 1;
 	}
-- 
Peter Stephenson <p.w.stephenson@ntlworld.com>
Web page now at http://homepage.ntlworld.com/p.w.stephenson/