New problem with ERR_RETURN ?

New problem with ERR_RETURN ?

[Bart Schaefer]

After zsh-workers 51001, 51071, 51076 (Philippe A.)

This seems wrong:

% setopt errreturn
% f() {
    while return 5
    do
        echo fail while1
        break
    done
}
% f ; echo $?
%

Note that "echo $?" was not executed when it appears in the same
command line as "f".  I don't believe ERR_RETURN should be having an
effect there.

Compare:

% f
% echo $?
5
%

Re: New problem with ERR_RETURN ?

[Peter Stephenson]

> On 07/12/2022 16:17 Bart Schaefer <schaefer@brasslantern.com> wrote:
> After zsh-workers 51001, 51071, 51076 (Philippe A.)
> 
> This seems wrong:
> 
> % setopt errreturn
> % f() {
>     while return 5
>     do
>         echo fail while1
>         break
>     done
> }
> % f ; echo $?
> %
> 
> Note that "echo $?" was not executed when it appears in the same
> command line as "f".

I haven't had a chance to look but I'm guessing this is something
along the lines of not properly restoring the "breaks" variable when
we return from the function, or some equivalent lack of care
with "retflags".

pws

Re: New problem with ERR_RETURN ?

[Philippe Altherr]

[-- Attachment #1: Type: text/plain, Size: 1030 bytes --]

I don't think there is a problem. At least not a new problem because Zsh
5.8 and Zsh dev behave the same way.

Note that if you replace the body of "f" with just "false" you get the same
results.

Note also that if you run "return; echo $?" nothing is printed.

From that I concluded that the example behaves correctly. The function "f"
returns a non-zero status. Therefore the call to "f" triggers a return,
which at the top-level simply exits the script/returns to the command line
and thus bypasses what comes next.

Philippe


On Wed, Dec 7, 2022 at 5:17 PM Bart Schaefer <schaefer@brasslantern.com>
wrote:

> After zsh-workers 51001, 51071, 51076 (Philippe A.)
>
> This seems wrong:
>
> % setopt errreturn
> % f() {
>     while return 5
>     do
>         echo fail while1
>         break
>     done
> }
> % f ; echo $?
> %
>
> Note that "echo $?" was not executed when it appears in the same
> command line as "f".  I don't believe ERR_RETURN should be having an
> effect there.
>
> Compare:
>
> % f
> % echo $?
> 5
> %
>
>

[-- Attachment #2: Type: text/html, Size: 1602 bytes --]

Re: New problem with ERR_RETURN ?

[Peter Stephenson]

> On 07/12/2022 16:41 Philippe Altherr <philippe.altherr@gmail.com> wrote:
> I don't think there is a problem. At least not a new problem because Zsh 5.8 and Zsh dev behave the same way.
> 
> Note that if you replace the body of "f" with just "false" you get the same results.

Yes, you're correct.  We should probably document this somehow.

pws

diff --git a/Doc/Zsh/options.yo b/Doc/Zsh/options.yo
index 445052617..dbaab08f4 100644
--- a/Doc/Zsh/options.yo
+++ b/Doc/Zsh/options.yo
@@ -1744,6 +1744,12 @@ Exiting due to tt(ERR_EXIT) has certain interactions with asynchronous
 jobs noted in
 ifzman(the section JOBS in zmanref(zshmisc))\
 ifnzman(noderef(Jobs & Signals)).
+
+It should also be noted that this option forces an immediate
+return to the command prompt when the non-zero status occurs
+in a sequence of commands typed on the command line.  In other
+words, the sequence of commands typed by the user may be
+thought of as a script for this purpose.
 )
 pindex(ERR_RETURN)
 pindex(NO_ERR_RETURN)
@@ -1772,6 +1778,9 @@ function, code inside it is considered separately: it may force a return
 from tt(summit) (assuming the option remains set within tt(summit)), but
 not from the enclosing context.  This behaviour is different from
 tt(ERR_EXIT) which is unaffected by function scope.
+
+Like tt(ERR_EXIT), this option forces an immediate return to the
+command prompt in interactive shells.
 )
 pindex(EVAL_LINENO)
 pindex(NO_EVAL_LINENO)

Re: New problem with ERR_RETURN ?

[Bart Schaefer]

On Wed, Dec 7, 2022 at 8:55 AM Peter Stephenson
<p.w.stephenson@ntlworld.com> wrote:
>
> +
> +It should also be noted that this option forces an immediate
> +return to the command prompt when the non-zero status occurs
> +in a sequence of commands typed on the command line.  In other
> +words, the sequence of commands typed by the user may be
> +thought of as a script for this purpose.

Um, no, it doesn't return to the prompt, it kills the shell.

Re: New problem with ERR_RETURN ?

[Peter Stephenson]

> On 07/12/2022 17:02 Bart Schaefer <schaefer@brasslantern.com> wrote:
> On Wed, Dec 7, 2022 at 8:55 AM Peter Stephenson
> <p.w.stephenson@ntlworld.com> wrote:
> >
> > +
> > +It should also be noted that this option forces an immediate
> > +return to the command prompt when the non-zero status occurs
> > +in a sequence of commands typed on the command line.  In other
> > +words, the sequence of commands typed by the user may be
> > +thought of as a script for this purpose.
> 
> Um, no, it doesn't return to the prompt, it kills the shell.

Yes, that would presumably require testing with a different
option, wouldn't it?

diff --git a/Doc/Zsh/options.yo b/Doc/Zsh/options.yo
index 445052617..b2213eb44 100644
--- a/Doc/Zsh/options.yo
+++ b/Doc/Zsh/options.yo
@@ -1744,6 +1744,9 @@ Exiting due to tt(ERR_EXIT) has certain interactions with asynchronous
 jobs noted in
 ifzman(the section JOBS in zmanref(zshmisc))\
 ifnzman(noderef(Jobs & Signals)).
+
+Note this behaviour is not disabled in interactive shells ---
+a non-zero status on the command line causes the shell to exit.
 )
 pindex(ERR_RETURN)
 pindex(NO_ERR_RETURN)
@@ -1772,6 +1775,12 @@ function, code inside it is considered separately: it may force a return
 from tt(summit) (assuming the option remains set within tt(summit)), but
 not from the enclosing context.  This behaviour is different from
 tt(ERR_EXIT) which is unaffected by function scope.
+
+It should also be noted that this option forces an immediate
+return to the command prompt when the non-zero status occurs
+in a sequence of commands typed on the command line.  In other
+words, the sequence of commands typed by the user may be
+thought of as a function for this purpose.
 )
 pindex(EVAL_LINENO)
 pindex(NO_EVAL_LINENO)

Re: New problem with ERR_RETURN ?

[Lawrence Velázquez]

On Wed, Dec 7, 2022, at 12:07 PM, Peter Stephenson wrote:
> @@ -1772,6 +1775,12 @@ function, code inside it is considered 
> separately: it may force a return
>  from tt(summit) (assuming the option remains set within tt(summit)), 
> but
>  not from the enclosing context.  This behaviour is different from
>  tt(ERR_EXIT) which is unaffected by function scope.
> +
> +It should also be noted that this option forces an immediate
> +return to the command prompt when the non-zero status occurs
> +in a sequence of commands typed on the command line.  In other
> +words, the sequence of commands typed by the user may be
> +thought of as a function for this purpose.
>  )
>  pindex(EVAL_LINENO)
>  pindex(NO_EVAL_LINENO)

Bikeshedding perhaps, but this sentence already exists just above:

	This will trigger an exit at the outermost level of a
	non-interactive script.

Maybe any new text should come right after this?  That would keep
the explanations about how ERR_RETURN affects behavior at the top
level in one place.

-- 
vq

Re: New problem with ERR_RETURN ?

[Peter Stephenson]

> On 08/12/2022 06:05 Lawrence Velázquez <larryv@zsh.org> wrote:
> Bikeshedding perhaps, but this sentence already exists just above:
> 
> 	This will trigger an exit at the outermost level of a
> 	non-interactive script.
> 
> Maybe any new text should come right after this?  That would keep
> the explanations about how ERR_RETURN affects behavior at the top
> level in one place.

That will make it more compact, too.

pws

diff --git a/Doc/Zsh/options.yo b/Doc/Zsh/options.yo
index 445052617..e92969531 100644
--- a/Doc/Zsh/options.yo
+++ b/Doc/Zsh/options.yo
@@ -1744,6 +1744,9 @@ Exiting due to tt(ERR_EXIT) has certain interactions with asynchronous
 jobs noted in
 ifzman(the section JOBS in zmanref(zshmisc))\
 ifnzman(noderef(Jobs & Signals)).
+
+Note this behaviour is not disabled in interactive shells ---
+a non-zero status on the command line causes the shell to exit.
 )
 pindex(ERR_RETURN)
 pindex(NO_ERR_RETURN)
@@ -1756,7 +1759,10 @@ If a command has a non-zero exit status, return immediately from the
 enclosing function.  The logic is similar to that for tt(ERR_EXIT),
 except that an implicit tt(return) statement is executed instead of an
 tt(exit).  This will trigger an exit at the outermost level of a
-non-interactive script.
+non-interactive script.  At the top level of an interactive shell,
+it will trigger a return to the command prompt; in other
+words, the sequence of commands typed by the user may be
+thought of as a function for this purpose.
 
 Normally this option inherits the behaviour of tt(ERR_EXIT) that
 code followed by `tt(&&)' `tt(||)' does not trigger a return.  Hence

Re: New problem with ERR_RETURN ?

[Bart Schaefer]

On Wed, Dec 7, 2022 at 8:42 AM Philippe Altherr
<philippe.altherr@gmail.com> wrote:
>
> At least not a new problem because Zsh 5.8 and Zsh dev behave the same way.

I should have checked that.  Thanks.

> Note also that if you run "return; echo $?" nothing is printed.

That's "documented" in the code, though, as an explicit exception for
interactive mode.