Problem when run-hook-with-args with closure

Problem when run-hook-with-args with closure

[Zhongwei Yao]

Hi, list,
I try following code on both emacs 24.4.1 and 24.3.1:

(setq lexical-binding t)
(let ((x 1))
  (setq foo
        (lambda (y)
          (message "x+y=%d" (+ x y)))))
(run-hook-with-args 'foo 1)

On emacs 24.4.1, it runs and print: "x+y=2" as expected.
On emacs 24.3.1, it gives error:
  run-hook-with-args: Symbol's function definition is void: closure

By the way, this problem is from the compile.el and when
compilation-finish-functions is set, emacs 24.3.1 would give same error:
  Symbol's function definition is void: closure

And 24.4.1 is OK.

My problem is Why 24.3.1 returns this? How can I solve it on 24.3.1?

Thanks,
Zhongwei

Re: Problem when run-hook-with-args with closure

[Michael Heerdegen]

Zhongwei Yao <ashi08104@gmail.com> writes:

> Hi, list,
> I try following code on both emacs 24.4.1 and 24.3.1:
>
> (setq lexical-binding t)
> (let ((x 1))
>   (setq foo
>         (lambda (y)
>           (message "x+y=%d" (+ x y)))))
> (run-hook-with-args 'foo 1)
>
> On emacs 24.4.1, it runs and print: "x+y=2" as expected.
> On emacs 24.3.1, it gives error:
>   run-hook-with-args: Symbol's function definition is void: closure

That had been a bug in 24.3 IMO.

> And 24.4.1 is OK.

I also can't reproduce it using my trunk built. So it has seemingly be
fixed.

> How can I solve it on 24.3.1?

I can't give an advice better than "try to avoid the problem" or "please
upgrade".


Michael.

Re: Problem when run-hook-with-args with closure

[Zhongwei Yao]

On Fri, Dec 19, 2014 at 5:09 PM, Michael Heerdegen <michael_heerdegen@web.de
> wrote:
>
> Zhongwei Yao <ashi08104@gmail.com> writes:
>
> > Hi, list,
> > I try following code on both emacs 24.4.1 and 24.3.1:
> >
> > (setq lexical-binding t)
> > (let ((x 1))
> >   (setq foo
> >         (lambda (y)
> >           (message "x+y=%d" (+ x y)))))
> > (run-hook-with-args 'foo 1)
> >
> > On emacs 24.4.1, it runs and print: "x+y=2" as expected.
> > On emacs 24.3.1, it gives error:
> >   run-hook-with-args: Symbol's function definition is void: closure
>
> That had been a bug in 24.3 IMO.
>
> > And 24.4.1 is OK.
>
> I also can't reproduce it using my trunk built. So it has seemingly be
> fixed.
>
> > How can I solve it on 24.3.1?
>
> I can't give an advice better than "try to avoid the problem" or "please
> upgrade".
>
>
> Michael.
>
>
> Michael, Thanks.

However, a problem (http://sourceforge.net/p/cedet/mailman/message/33153606/)
in latest emacs (24.4.1) prevents me upgrading:(

--
Zhongwei

Re: Problem when run-hook-with-args with closure

[Michael Heerdegen]

Zhongwei Yao <ashi08104@gmail.com> writes:

> However, a problem
> (http://sourceforge.net/p/cedet/mailman/message/33153606/)
> in latest emacs (24.4.1) prevents me upgrading:(

I see. Have you made a bug report about that issue?

I don't use the related packages.  Looking at the code, evaluating
something like this might help as a workaround:

(defalias 'hif-equal (hif-mathify-binop equal))

Re: Problem when run-hook-with-args with closure

[Michael Heerdegen]

Zhongwei Yao <ashi08104@gmail.com> writes:

> Hi, list,
> I try following code on both emacs 24.4.1 and 24.3.1:
>
> (setq lexical-binding t)
> (let ((x 1))
>   (setq foo
>         (lambda (y)
>           (message "x+y=%d" (+ x y)))))
> (run-hook-with-args 'foo 1)
>
> On emacs 24.4.1, it runs and print: "x+y=2" as expected.
> On emacs 24.3.1, it gives error:
>   run-hook-with-args: Symbol's function definition is void: closure

That had been a bug in 24.3 IMO.

> And 24.4.1 is OK.

I also can't reproduce it using my trunk built. So it has seemingly be
fixed.

> How can I solve it on 24.3.1?

I can't give an advice better than "try to avoid the problem" or "please
upgrade".


Michael.

Re: Problem when run-hook-with-args with closure

[Stefan Monnier]

>   (setq foo
>         (lambda (y)
>           (message "x+y=%d" (+ x y)))))
> (run-hook-with-args 'foo 1)

Hooks should contain *lists* of functions.  The fact that they can also
just contain a function is an old backward compatibility hack: better
not rely on it.

So, if you do it right, you won't hit the bug:

     (setq foo
           (list (lambda (y)
                   (message "x+y=%d" (+ x y))))))
   (run-hook-with-args 'foo 1)

Or better yet:

     (add-hook 'foo
               (lambda (y)
                   (message "x+y=%d" (+ x y))))))
   (run-hook-with-args 'foo 1)


-- Stefan

Re: Problem when run-hook-with-args with closure

[Zhongwei Yao]

Hi, Stefan,
Thanks, it works!

Best Regards,
Zhongwei

On Fri, Dec 19, 2014 at 10:11 PM, Stefan Monnier <monnier@iro.umontreal.ca>
wrote:

> >   (setq foo
> >         (lambda (y)
> >           (message "x+y=%d" (+ x y)))))
> > (run-hook-with-args 'foo 1)
>
> Hooks should contain *lists* of functions.  The fact that they can also
> just contain a function is an old backward compatibility hack: better
> not rely on it.
>
> So, if you do it right, you won't hit the bug:
>
>      (setq foo
>            (list (lambda (y)
>                    (message "x+y=%d" (+ x y))))))
>    (run-hook-with-args 'foo 1)
>
> Or better yet:
>
>      (add-hook 'foo
>                (lambda (y)
>                    (message "x+y=%d" (+ x y))))))
>    (run-hook-with-args 'foo 1)
>
>
> -- Stefan
>
>
>

incremental search for "OK " matches "OK" followed by ^M for CR !?

[ISHIKAWA,Chiaki]

Hi,

With the latest released emacs,
I noticed that when editing a file with mixed line ending of LF-only, 
and CR-LF variety (I am editing a file that contains an excerpt from
a terminal session recorded by "script" command. These recorded lines 
are shown as ending with ^M) ,
incremental search, C-S, for "OK " matches
the string "OK" that is followed by CR which is shown as ^M.

That is, "OK " seems to match
literal "OK " and also "OK" at the end of a line with CRLF ending.
I was surprised to find the "OK^M" as being high-lighted (as possible 
matches are these days with font lock or whaever).

I don't think this was the case before, and the extra match
with "OK^M" interfered my search for a meaningful message line that 
contains literal "ok ".

Anyone noticed this?

TIA

RE: incremental search for "OK " matches "OK" followed by ^M for CR !?

[Drew Adams]

> I don't think this was the case before, and the extra match
> with "OK^M" interfered my search for a meaningful message line that
> contains literal "ok ".
> 
> Anyone noticed this?

Yes, the behavior was changed. See option `search-whitespace-regexp'.
Set it to nil to get rid of the new behavior. Or just toggle the
behavior with `M-s SPC'.

And see the Emacs manual, node `Special Search'.