GNU bug report logs - #19066
25.0.50; nadvice and interactive-form

Previous Next

To reply to this bug, email your comments to 19066 AT debbugs.gnu.org.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Leo Liu <sdl.web <at> gmail.com>
To: bug-gnu-emacs <at> gnu.org
Subject: 25.0.50; nadvice and interactive-form
Date: Sun, 16 Nov 2014 18:00:21 +0800

1. Start emacs and load a file with these lines:

--8<---------------cut here---------------start------------->8---
   (advice-add 'just-one-space :before #'change-interactive-form)
   (defun change-interactive-form (&optional x y)
     (interactive (list 1 1))
     (princ (list x y)))
--8<---------------cut here---------------end--------------->8---

M-x just-one-space prints (1 nil)

2. Start emacs and load a file with these lines:

--8<---------------cut here---------------start------------->8---
   (defun change-interactive-form (&optional x y)
     (interactive (list 1 1))
     (princ (list x y)))
   (advice-add 'just-one-space :before #'change-interactive-form)
--8<---------------cut here---------------end--------------->8---

M-x just-one-space errs: apply: Wrong number of arguments: (0 . 1), 2

Leo

Message #12 received at 19066 <at> debbugs.gnu.org (full text, mbox):

From: Lars Ingebrigtsen <larsi <at> gnus.org>
To: Leo Liu <sdl.web <at> gmail.com>
Cc: 19066 <at> debbugs.gnu.org, Stefan Monnier <monnier <at> iro.umontreal.ca>
Subject: Re: bug#19066: 25.0.50; nadvice and interactive-form
Date: Fri, 28 May 2021 04:11:13 +0200

Leo Liu <sdl.web <at> gmail.com> writes:

> 1. Start emacs and load a file with these lines:
>
>    (advice-add 'just-one-space :before #'change-interactive-form)
>    (defun change-interactive-form (&optional x y)
>      (interactive (list 1 1))
>      (princ (list x y)))
>
> M-x just-one-space prints (1 nil)
>
> 2. Start emacs and load a file with these lines:
>
>    (defun change-interactive-form (&optional x y)
>      (interactive (list 1 1))
>      (princ (list x y)))
>    (advice-add 'just-one-space :before #'change-interactive-form)
>
> M-x just-one-space errs: apply: Wrong number of arguments: (0 . 1), 2

So whether this works depends on whether `change-interactive-form' is
defined or not when `advice-add' is run?  (The only difference here is
the order.)

Perhaps Stefan has some insights here; added to the CCs.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Message #15 received at 19066 <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Lars Ingebrigtsen <larsi <at> gnus.org>
Cc: Leo Liu <sdl.web <at> gmail.com>, 19066 <at> debbugs.gnu.org
Subject: Re: bug#19066: 25.0.50; nadvice and interactive-form
Date: Sun, 30 May 2021 13:26:52 -0400

Lars Ingebrigtsen [2021-05-28 04:11:13] wrote:
> Leo Liu <sdl.web <at> gmail.com> writes:
>> 1. Start emacs and load a file with these lines:
>>
>>    (advice-add 'just-one-space :before #'change-interactive-form)
>>    (defun change-interactive-form (&optional x y)
>>      (interactive (list 1 1))
>>      (princ (list x y)))
>>
>> M-x just-one-space prints (1 nil)
>>
>> 2. Start emacs and load a file with these lines:
>>
>>    (defun change-interactive-form (&optional x y)
>>      (interactive (list 1 1))
>>      (princ (list x y)))
>>    (advice-add 'just-one-space :before #'change-interactive-form)
>>
>> M-x just-one-space errs: apply: Wrong number of arguments: (0 . 1), 2
>
> So whether this works depends on whether `change-interactive-form' is
> defined or not when `advice-add' is run?  (The only difference here is
> the order.)
> Perhaps Stefan has some insights here; added to the CCs.

[ Side note: the correct/desired behavior is the one described by (2),
  i.e. it should signal a "Wrong number of arguments" (tho after
  printing (1 1)).  ]

`advice-add` creates a new function object which is a "composition" of
the old function and the advice that's added to it.

In the case where `change-interactive-form` is defined later, one way to
make this work right would be for `advice-add` to put some kind of
"watcher" on (symbol-function 'change-interactive-form) to recompute the
interactive form of the "composed function".

Another way would be to make the interactive form of the "compose
function" more dynamic: make it re-checks the interactive form of every
advice function every time.  That's probably the better option.


        Stefan

Message #18 received at 19066 <at> debbugs.gnu.org (full text, mbox):

From: Lars Ingebrigtsen <larsi <at> gnus.org>
To: Stefan Monnier <monnier <at> iro.umontreal.ca>
Cc: Leo Liu <sdl.web <at> gmail.com>, 19066 <at> debbugs.gnu.org
Subject: Re: bug#19066: 25.0.50; nadvice and interactive-form
Date: Mon, 31 May 2021 07:43:34 +0200

Stefan Monnier <monnier <at> iro.umontreal.ca> writes:

> Another way would be to make the interactive form of the "compose
> function" more dynamic: make it re-checks the interactive form of every
> advice function every time.  That's probably the better option.

That sounds like a good solution to me -- the interactive path isn't
normally performance critical, so adding more work in that case should
be fine.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Message #21 received at 19066 <at> debbugs.gnu.org (full text, mbox):

From: Stefan Monnier <monnier <at> iro.umontreal.ca>
To: Lars Ingebrigtsen <larsi <at> gnus.org>
Cc: Leo Liu <sdl.web <at> gmail.com>, 19066 <at> debbugs.gnu.org
Subject: Re: bug#19066: 25.0.50; nadvice and interactive-form
Date: Mon, 31 May 2021 08:49:27 -0400

Lars Ingebrigtsen [2021-05-31 07:43:34] wrote:
> Stefan Monnier <monnier <at> iro.umontreal.ca> writes:
>> Another way would be to make the interactive form of the "compose
>> function" more dynamic: make it re-checks the interactive form of every
>> advice function every time.  That's probably the better option.
> That sounds like a good solution to me -- the interactive path isn't
> normally performance critical, so adding more work in that case should
> be fine.

Yes, performance is not an issue.  But preserving the semantics 100% is
currently somewhere between hard and impossible.  99.9% is fairly easy,
OTOH (IIRC the main problem has to do with how the args get represented
in `command-history`).


        Stefan

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.