4.2.0-pre-4

4.2.0-pre-4

[Peter Stephenson]

I've uploaded 4.2.0-pre-4 to ftp.zsh.org.

Oliver has got HP-UX 11 working with help from Paul Ackersviller.  This may
not be quite the end of that saga, but I wanted to get something out
before the weekend.  The zpty module is still not working, although it
might now be something quite simple.

For AIX we need to make sure any functions needed in libraries have
`mod_export' in front... Ibraheem Umaru-Mohammed has been providing
information on failures; there's an extra symbol this time.  There may
be more missing, however.

It ought to be possible to use `nm' on some other system to verify the
symbols that need exporting, and hence keep the list up to date, but I
don't think I'm going to have time for that.  With a bit of perl this
shouldn't be too hard a project.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************

Re: 4.2.0-pre-4

[Peter Stephenson]

Peter Stephenson wrote:
> It ought to be possible to use `nm' on some other system to verify the
> symbols that need exporting, and hence keep the list up to date, but I
> don't think I'm going to have time for that.

I lied.  Should have waited to release 4.2.0-pre-4.  This patch (the
first two hunks) will probably be needed on AIX.  Sent to zsh-users in
case it affects anybody downloading that.

(I'm assuming that stdout@@GLIBC_2.0 don't need exporting...)

Index: Src/Zle/zle_misc.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Zle/zle_misc.c,v
retrieving revision 1.10
diff -u -r1.10 zle_misc.c
--- Src/Zle/zle_misc.c	8 Mar 2004 11:44:14 -0000	1.10
+++ Src/Zle/zle_misc.c	12 Mar 2004 18:40:07 -0000
@@ -73,7 +73,7 @@
 }
 
 /**/
-int
+mod_export int
 selfinsertunmeta(char **args)
 {
     lastchar &= 0x7f;
Index: Src/Zle/zle_utils.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Zle/zle_utils.c,v
retrieving revision 1.8
diff -u -r1.8 zle_utils.c
--- Src/Zle/zle_utils.c	29 Oct 2003 19:17:48 -0000	1.8
+++ Src/Zle/zle_utils.c	12 Mar 2004 18:40:07 -0000
@@ -388,7 +388,7 @@
  * The message must be metafied.                              */
 
 /**/
-void mod_export
+mod_export void
 showmsg(char const *msg)
 {
     char const *p;
Index: Util/check_exports
===================================================================
RCS file: Util/check_exports
diff -N Util/check_exports
--- /dev/null	1 Jan 1970 00:00:00 -0000
+++ Util/check_exports	12 Mar 2004 18:40:07 -0000
@@ -0,0 +1,64 @@
+#!/usr/local/bin/perl -w
+
+# Attempt to scan executable, libraries, and .export files after
+# a zsh build to see if all necessary symbols appear in the .export file
+# (and hence with `mod_export' in the source file).  This keeps AIX happy.
+# Probably severely system dependent, but known to run on Fedora Core 1,
+# at least.  Not needed on AIX itself... you can tell if doesn't link.
+
+if (! -f "zsh") {
+    die "Can't file zsh, are we in the Src directory of the build?\n";
+}
+
+my (%defined, %undefined, %exported);
+
+foreach my $file ("zsh", glob("*.so */*.so")) {
+    next unless -f $file;
+
+    my $exports = $file;
+    $exports =~ s/\.so//;
+    $exports .= ".export";
+    if (-f $exports) {
+	open EXPORT, $exports  or  die "Can't read $exports: $!\n";
+	my $href = $exported{$file} = { };
+	while (<EXPORT>) {
+	    next if /^#/;
+	    chomp;
+	    $href->{$_} = 1;
+	}
+	close EXPORT;
+    } else {
+	warn "Hmmm... no .exports file for $file\n";
+    }
+
+    open PIPE, "nm $file |"  or  die "Can't popen nm";
+    while (<PIPE>) {
+	s/^[0-9a-f]*\s+//;
+	my ($type, $sym) = split;
+	# ignore local symbols (lower case)
+	if ($type =~ /^[TBAD]/) {
+	    if (!defined $defined{$sym}) {
+		$defined{$sym} = $file;
+	    }
+	} elsif ($type eq 'U') {
+	    # could skip undefined from zsh and zsh.so, but what the heck
+	    my $ref = \$undefined{$sym};
+	    if (defined $$ref) {
+		push @$$ref, $file;
+	    } else {
+		$$ref = [ $file ];
+	    }
+	}
+    }
+    close PIPE  or  die "nm failed";
+}
+
+foreach $sym (keys %undefined) {
+    my $deffile = $defined{$sym};
+    if (defined $deffile) {
+	if (!$exported{$deffile}{$sym}) {
+	    printf "%-20s: %-20s: %s\n", $sym, $defined{$sym},
+	    join(" ", @{$undefined{$sym}});
+	}
+    }
+}

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************

Re: 4.2.0-pre-4

[Ibraheem Umaru-Mohammed]

On Fri, 2004-03-12 at 18:58, Peter Stephenson wrote:
> Peter Stephenson wrote:
> > It ought to be possible to use `nm' on some other system to verify the
> > symbols that need exporting, and hence keep the list up to date, but I
> > don't think I'm going to have time for that.
> 
> I lied.  Should have waited to release 4.2.0-pre-4.  This patch (the
> first two hunks) will probably be needed on AIX.  Sent to zsh-users in
> case it affects anybody downloading that.
> 
> (I'm assuming that stdout@@GLIBC_2.0 don't need exporting...)
> 
> Index: Src/Zle/zle_misc.c
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Src/Zle/zle_misc.c,v
> retrieving revision 1.10
> diff -u -r1.10 zle_misc.c
> --- Src/Zle/zle_misc.c	8 Mar 2004 11:44:14 -0000	1.10
> +++ Src/Zle/zle_misc.c	12 Mar 2004 18:40:07 -0000
> @@ -73,7 +73,7 @@
>  }
>  
>  /**/
> -int
> +mod_export int
>  selfinsertunmeta(char **args)
>  {
>      lastchar &= 0x7f;
> Index: Src/Zle/zle_utils.c
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Src/Zle/zle_utils.c,v
> retrieving revision 1.8
> diff -u -r1.8 zle_utils.c
> --- Src/Zle/zle_utils.c	29 Oct 2003 19:17:48 -0000	1.8
> +++ Src/Zle/zle_utils.c	12 Mar 2004 18:40:07 -0000
> @@ -388,7 +388,7 @@
>   * The message must be metafied.                              */
>  
>  /**/
> -void mod_export
> +mod_export void
>  showmsg(char const *msg)
>  {
>      char const *p;
> Index: Util/check_exports
> ===================================================================
> RCS file: Util/check_exports
> diff -N Util/check_exports
> --- /dev/null	1 Jan 1970 00:00:00 -0000
> +++ Util/check_exports	12 Mar 2004 18:40:07 -0000
> @@ -0,0 +1,64 @@
> +#!/usr/local/bin/perl -w
> +
> +# Attempt to scan executable, libraries, and .export files after
> +# a zsh build to see if all necessary symbols appear in the .export file
> +# (and hence with `mod_export' in the source file).  This keeps AIX happy.
> +# Probably severely system dependent, but known to run on Fedora Core 1,
> +# at least.  Not needed on AIX itself... you can tell if doesn't link.
> +
> +if (! -f "zsh") {
> +    die "Can't file zsh, are we in the Src directory of the build?\n";
> +}
> +
> +my (%defined, %undefined, %exported);
> +
> +foreach my $file ("zsh", glob("*.so */*.so")) {
> +    next unless -f $file;
> +
> +    my $exports = $file;
> +    $exports =~ s/\.so//;
> +    $exports .= ".export";
> +    if (-f $exports) {
> +	open EXPORT, $exports  or  die "Can't read $exports: $!\n";
> +	my $href = $exported{$file} = { };
> +	while (<EXPORT>) {
> +	    next if /^#/;
> +	    chomp;
> +	    $href->{$_} = 1;
> +	}
> +	close EXPORT;
> +    } else {
> +	warn "Hmmm... no .exports file for $file\n";
> +    }
> +
> +    open PIPE, "nm $file |"  or  die "Can't popen nm";
> +    while (<PIPE>) {
> +	s/^[0-9a-f]*\s+//;
> +	my ($type, $sym) = split;
> +	# ignore local symbols (lower case)
> +	if ($type =~ /^[TBAD]/) {
> +	    if (!defined $defined{$sym}) {
> +		$defined{$sym} = $file;
> +	    }
> +	} elsif ($type eq 'U') {
> +	    # could skip undefined from zsh and zsh.so, but what the heck
> +	    my $ref = \$undefined{$sym};
> +	    if (defined $$ref) {
> +		push @$$ref, $file;
> +	    } else {
> +		$$ref = [ $file ];
> +	    }
> +	}
> +    }
> +    close PIPE  or  die "nm failed";
> +}
> +
> +foreach $sym (keys %undefined) {
> +    my $deffile = $defined{$sym};
> +    if (defined $deffile) {
> +	if (!$exported{$deffile}{$sym}) {
> +	    printf "%-20s: %-20s: %s\n", $sym, $defined{$sym},
> +	    join(" ", @{$undefined{$sym}});
> +	}
> +    }
> +}

Peter,

It has all finally compiled on AIX 5.2, and all seems to work correctly.
Thanks for taking the time to look at this - much appreciated.

Cheers,

			--ibraheem. 

Re: 4.2.0-pre-4

[Peter Stephenson]

Ibraheem Umaru-Mohammed wrote:
> It has all finally compiled on AIX 5.2, and all seems to work correctly.
> Thanks for taking the time to look at this - much appreciated.

This is good, thanks for testing... as far as I know zsh compiles
everywhere (for suitable values of `everywhere'), though with a few
remaining zpty problems.  It would be nice to fix those, but it's not a
top priority any more...  I will release 4.2.0 on Friday anyway, unless
something major turns up.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK                          Tel: +44 (0)1223 692070


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************

zle_thingy.c: variable "modsave" used before its value set

[Jens Petersen]

Hi,

I received a bug report that the local variable `modsave' is
used uninitialised in bin_zle_call() (zle_thingy.c).

	https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=116561

Is this something not worth losing any sleep over, or what
should it be initialised too?

Jens

Re: zle_thingy.c: variable "modsave" used before its value set

[Wayne Davison]

[-- Attachment #1: Type: text/plain, Size: 525 bytes --]

On Mon, Apr 12, 2004 at 10:38:00AM +0900, Jens Petersen wrote:
> I received a bug report that the local variable `modsave' is
> used uninitialised in bin_zle_call() (zle_thingy.c).

The variable can't actually be used uninitialized -- the compiler just
thinks it can.

If we want to get rid of the warning, I'd recommend making the struct
static rather than initializing it to an unneeded value on every call.
Here's a patch that does this, plus it gets rid of some unneeded cleanup
code that can't get triggered.

..wayne..

[-- Attachment #2: modsave.patch --]
[-- Type: text/plain, Size: 496 bytes --]

--- Src/Zle/zle_thingy.c	12 Dec 2003 22:53:28 -0000	1.12
+++ Src/Zle/zle_thingy.c	12 Apr 2004 08:40:04 -0000
@@ -622,13 +622,11 @@ static int
 bin_zle_call(char *name, char **args, Options ops, char func)
 {
     Thingy t;
-    struct modifier modsave;
+    static struct modifier modsave;
     int ret, saveflag = 0;
     char *wname = *args++;
 
     if (!wname) {
-	if (saveflag)
-	    zmod = modsave;
 	return (!zleactive || incompctlfunc || incompfunc ||
 		sfcontext != SFC_WIDGET);
     }

Re: zle_thingy.c: variable "modsave" used before its value set

[Bart Schaefer]

On Apr 12,  2:04am, Wayne Davison wrote:
> 
> The variable can't actually be used uninitialized -- the compiler just
> thinks it can.
> 
> If we want to get rid of the warning, I'd recommend making the struct
> static rather than initializing it to an unneeded value on every call.

Are you really sure that bin_zle_call() is not re-entrant?  Seems to me
that the whole point of modsave is that it is a stack variable.

Re: zle_thingy.c: variable "modsave" used before its value set

[Wayne Davison]

On Mon, Apr 12, 2004 at 08:21:20AM -0700, Bart Schaefer wrote:
> Are you really sure that bin_zle_call() is not re-entrant?

No, I'm not, so that static idea wouldn't work.  The useless
cleanup code could be safely removed, though, if desired.

..wayne..