Should declare -p add a new declaration inside a function?

Should declare -p add a new declaration inside a function?

[Rocky Bernstein]

[-- Attachment #1: Type: text/plain, Size: 621 bytes --]

It appears that using "declare" or "typeset" with the -p (print) flag will
add a new declaration inside a function. Is this really desirable? It seems
to reduce the usefulness of -p.

Here's an example:

declare -a xx
xx=(an array)
bug() {
    echo xx is \"${xx[@]}\" before \"declare -p xx\"
    echo 'declare -p xx:'
    declare -p xx
    echo xx is now \"${xx[@]}\" in bug
    echo '===='
}
bug
echo 'declare -p xx outside bug:'
declare -p xx


Running this produces:

xx is "an array" before "declare -p xx"
declare -p xx:
typeset xx=''
xx is now "" in bug
====
declare -p xx outside bug:
typeset -a xx
xx=(an array)

[-- Attachment #2: Type: text/html, Size: 933 bytes --]

Re: Should declare -p add a new declaration inside a function?

[Peter Stephenson]

On Sat, 13 Sep 2008 10:52:35 -0400
"Rocky Bernstein" <rocky.bernstein@gmail.com> wrote:
> It appears that using "declare" or "typeset" with the -p (print) flag will
> add a new declaration inside a function. Is this really desirable? It seems
> to reduce the usefulness of -p.

That sure as heck looks like a bug to me.  It shouldn't ever create
anything, even if the variable doesn't exist; it's purely for
information.  I don't think "declare -p" has had much love and
attention; it doesn't look like it's ever worked with arguments
other than patterns (with -m).

Index: Src/builtin.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/builtin.c,v
retrieving revision 1.205
diff -u -r1.205 builtin.c
--- Src/builtin.c	11 Sep 2008 17:14:39 -0000	1.205
+++ Src/builtin.c	13 Sep 2008 19:49:33 -0000
@@ -2473,10 +2473,17 @@
 
     /* Take arguments literally.  Don't glob */
     while ((asg = getasg(*argv++))) {
-	if (!typeset_single(name, asg->name,
-			    (Param) (paramtab == realparamtab ?
-				     gethashnode2(paramtab, asg->name) :
-				     paramtab->getnode(paramtab, asg->name)),
+	HashNode hn = (paramtab == realparamtab ?
+		       gethashnode2(paramtab, asg->name) :
+		       paramtab->getnode(paramtab, asg->name));
+	if (OPT_ISSET(ops,'p')) {
+	    if (hn)
+		printparamnode(hn, printflags);
+	    else
+		zwarnnam(name, "no such variable: %s", asg->name);
+	    continue;
+	}
+	if (!typeset_single(name, asg->name, (Param)hn,
 			    func, on, off, roff, asg->value, NULL,
 			    ops, 0))
 	    returnval = 1;
Index: Test/B02typeset.ztst
===================================================================
RCS file: /cvsroot/zsh/zsh/Test/B02typeset.ztst,v
retrieving revision 1.18
diff -u -r1.18 B02typeset.ztst
--- Test/B02typeset.ztst	16 Dec 2007 14:05:16 -0000	1.18
+++ Test/B02typeset.ztst	13 Sep 2008 19:49:33 -0000
@@ -444,3 +444,12 @@
 0:Lower case conversion, does not apply to values used internally
 >lower
 >value of $lower
+
+ typeset -a array
+ array=(foo bar)
+ fn() { typeset -p array nonexistent; }
+ fn
+0:declare -p shouldn't create scoped values
+>typeset -a array
+>array=(foo bar)
+?fn:typeset: no such variable: nonexistent


-- 
Peter Stephenson <p.w.stephenson@ntlworld.com>
Web page now at http://homepage.ntlworld.com/p.w.stephenson/

Re: Should declare -p add a new declaration inside a function?

[Rocky Bernstein]

[-- Attachment #1: Type: text/plain, Size: 3318 bytes --]

Thanks.  This helps a lot.

A couple things to consider. Right now if the variable is not declared,
declare -p  returns 0. Should it be nonzero as is the case if you give a bad
option?

Is there a way to figure out or get a list of local variables versus
non-local variables versus global variables? By "local" I mean those that
were defined in the most recent scope and by "global" I mean those defined
outside of any function nesting, and by "non-local" I mean "not defined in
the immediate scope."

Thanks again

On Sat, Sep 13, 2008 at 3:55 PM, Peter Stephenson <
p.w.stephenson@ntlworld.com> wrote:

> On Sat, 13 Sep 2008 10:52:35 -0400
> "Rocky Bernstein" <rocky.bernstein@gmail.com> wrote:
> > It appears that using "declare" or "typeset" with the -p (print) flag
> will
> > add a new declaration inside a function. Is this really desirable? It
> seems
> > to reduce the usefulness of -p.
>
> That sure as heck looks like a bug to me.  It shouldn't ever create
> anything, even if the variable doesn't exist; it's purely for
> information.  I don't think "declare -p" has had much love and
> attention; it doesn't look like it's ever worked with arguments
> other than patterns (with -m).
>
> Index: Src/builtin.c
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Src/builtin.c,v
> retrieving revision 1.205
> diff -u -r1.205 builtin.c
> --- Src/builtin.c       11 Sep 2008 17:14:39 -0000      1.205
> +++ Src/builtin.c       13 Sep 2008 19:49:33 -0000
> @@ -2473,10 +2473,17 @@
>
>     /* Take arguments literally.  Don't glob */
>     while ((asg = getasg(*argv++))) {
> -       if (!typeset_single(name, asg->name,
> -                           (Param) (paramtab == realparamtab ?
> -                                    gethashnode2(paramtab, asg->name) :
> -                                    paramtab->getnode(paramtab,
> asg->name)),
> +       HashNode hn = (paramtab == realparamtab ?
> +                      gethashnode2(paramtab, asg->name) :
> +                      paramtab->getnode(paramtab, asg->name));
> +       if (OPT_ISSET(ops,'p')) {
> +           if (hn)
> +               printparamnode(hn, printflags);
> +           else
> +               zwarnnam(name, "no such variable: %s", asg->name);
> +           continue;
> +       }
> +       if (!typeset_single(name, asg->name, (Param)hn,
>                            func, on, off, roff, asg->value, NULL,
>                            ops, 0))
>            returnval = 1;
> Index: Test/B02typeset.ztst
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Test/B02typeset.ztst,v
> retrieving revision 1.18
> diff -u -r1.18 B02typeset.ztst
> --- Test/B02typeset.ztst        16 Dec 2007 14:05:16 -0000      1.18
> +++ Test/B02typeset.ztst        13 Sep 2008 19:49:33 -0000
> @@ -444,3 +444,12 @@
>  0:Lower case conversion, does not apply to values used internally
>  >lower
>  >value of $lower
> +
> + typeset -a array
> + array=(foo bar)
> + fn() { typeset -p array nonexistent; }
> + fn
> +0:declare -p shouldn't create scoped values
> +>typeset -a array
> +>array=(foo bar)
> +?fn:typeset: no such variable: nonexistent
>
>
> --
> Peter Stephenson <p.w.stephenson@ntlworld.com>
> Web page now at http://homepage.ntlworld.com/p.w.stephenson/
>

[-- Attachment #2: Type: text/html, Size: 5080 bytes --]

Re: Should declare -p add a new declaration inside a function?

[Peter Stephenson]

On Sat, 13 Sep 2008 20:48:56 -0400
"Rocky Bernstein" <rocky.bernstein@gmail.com> wrote:
> A couple things to consider. Right now if the variable is not declared,
> declare -p  returns 0. Should it be nonzero as is the case if you give a bad
> option?

Yes, I missed that the first time.

> Is there a way to figure out or get a list of local variables versus
> non-local variables versus global variables? By "local" I mean those that
> were defined in the most recent scope and by "global" I mean those defined
> outside of any function nesting, and by "non-local" I mean "not defined in
> the immediate scope."

You can use $parameters:

fn() {
  local foo
  print ${(k)parameters[(R)*local*]}
}

${(k)parameters[(R)^*local*]} (with extendedglob) gets you the global
parameters.  There's no way of telling which function level a variable is
local at.

Index: Src/builtin.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/builtin.c,v
retrieving revision 1.206
diff -u -r1.206 builtin.c
--- Src/builtin.c	13 Sep 2008 19:59:14 -0000	1.206
+++ Src/builtin.c	15 Sep 2008 08:52:40 -0000
@@ -2479,8 +2479,10 @@
 	if (OPT_ISSET(ops,'p')) {
 	    if (hn)
 		printparamnode(hn, printflags);
-	    else
+	    else {
 		zwarnnam(name, "no such variable: %s", asg->name);
+		returnval = 1;
+	    }
 	    continue;
 	}
 	if (!typeset_single(name, asg->name, (Param)hn,
Index: Test/B02typeset.ztst
===================================================================
RCS file: /cvsroot/zsh/zsh/Test/B02typeset.ztst,v
retrieving revision 1.19
diff -u -r1.19 B02typeset.ztst
--- Test/B02typeset.ztst	13 Sep 2008 19:59:14 -0000	1.19
+++ Test/B02typeset.ztst	15 Sep 2008 08:52:40 -0000
@@ -449,7 +449,7 @@
  array=(foo bar)
  fn() { typeset -p array nonexistent; }
  fn
-0:declare -p shouldn't create scoped values
+1:declare -p shouldn't create scoped values
 >typeset -a array
 >array=(foo bar)
 ?fn:typeset: no such variable: nonexistent


-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070

Re: Should declare -p add a new declaration inside a function?

[Rocky Bernstein]

[-- Attachment #1: Type: text/plain, Size: 2942 bytes --]

On Mon, Sep 15, 2008 at 4:53 AM, Peter Stephenson <pws@csr.com> wrote:

> On Sat, 13 Sep 2008 20:48:56 -0400
> "Rocky Bernstein" <rocky.bernstein@gmail.com> wrote:
> > A couple things to consider. Right now if the variable is not declared,
> > declare -p  returns 0. Should it be nonzero as is the case if you give a
> bad
> > option?
>
> Yes, I missed that the first time.


Thanks. As many times in the past, this  helps a lot. I had tried to parse
error output in a regular expression but that's problematic and not as
simple.


>
>
> > Is there a way to figure out or get a list of local variables versus
> > non-local variables versus global variables? By "local" I mean those that
> > were defined in the most recent scope and by "global" I mean those
> defined
> > outside of any function nesting, and by "non-local" I mean "not defined
> in
> > the immediate scope."
>
> You can use $parameters:
>
> fn() {
>  local foo
>  print ${(k)parameters[(R)*local*]}
> }


Interesting. This seems to show any variables defined local somewhere in the
call stack (not just the immediate one).



>
> ${(k)parameters[(R)^*local*]} (with extendedglob) gets you the global
> parameters.  There's no way of telling which function level a variable is
> local at.


There are a number of possibilities for tracking what variable goes with
what stack frame. Probably best to defer deciding on how exactly to deal
with.




>
>
> Index: Src/builtin.c
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Src/builtin.c,v
> retrieving revision 1.206
> diff -u -r1.206 builtin.c
> --- Src/builtin.c       13 Sep 2008 19:59:14 -0000      1.206
> +++ Src/builtin.c       15 Sep 2008 08:52:40 -0000
> @@ -2479,8 +2479,10 @@
>         if (OPT_ISSET(ops,'p')) {
>            if (hn)
>                printparamnode(hn, printflags);
> -           else
> +           else {
>                 zwarnnam(name, "no such variable: %s", asg->name);
> +               returnval = 1;
> +           }
>             continue;
>        }
>        if (!typeset_single(name, asg->name, (Param)hn,
> Index: Test/B02typeset.ztst
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Test/B02typeset.ztst,v
> retrieving revision 1.19
> diff -u -r1.19 B02typeset.ztst
> --- Test/B02typeset.ztst        13 Sep 2008 19:59:14 -0000      1.19
> +++ Test/B02typeset.ztst        15 Sep 2008 08:52:40 -0000
> @@ -449,7 +449,7 @@
>   array=(foo bar)
>  fn() { typeset -p array nonexistent; }
>  fn
> -0:declare -p shouldn't create scoped values
> +1:declare -p shouldn't create scoped values
>  >typeset -a array
>  >array=(foo bar)
>  ?fn:typeset: no such variable: nonexistent
>
>
> --
> Peter Stephenson <pws@csr.com>                  Software Engineer
> CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
> Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070
>

[-- Attachment #2: Type: text/html, Size: 4964 bytes --]