bug#20600: 24.5.3: add-hook - wrong arguments error

bug#20600: 24.5.3: add-hook - wrong arguments error

[Andreas Röhler]

[-- Attachment #1: Type: text/plain, Size: 559 bytes --]

||after-change-functions|  wrongly accepts functions with three args only
  
Issue is explained in docstring of workaround below.
If the function to be hooked in --save-buffer-- doesn't require args, add-hook resp. after-change-functions should not.

(defun my-instant-save-buffer (eins zwei drei)
   "To be hooked into list `after-change-functions'

`after-change-functions' expects functions receiving three arguments.
Arguments are ignored here, but slots needed by add-hook"
   (save-buffer))

(add-hook 'after-change-functions 'my-instant-save-buffer)|


[-- Attachment #2: Type: text/html, Size: 752 bytes --]

bug#20600: 24.5.3: add-hook - wrong arguments error

[Artur Malabarba]

[-- Attachment #1: Type: text/plain, Size: 249 bytes --]

The function you present is not a workaround, it's the right way to do this.

Still, it sounds like it might be useful for run-hooks to properly handle
functions whose arity is less than the expected arity.
Does anyone foresee any issues with that?

[-- Attachment #2: Type: text/html, Size: 290 bytes --]

bug#20600: 24.5.3: add-hook - wrong arguments error

[Eli Zaretskii]

> Date: Mon, 18 May 2015 12:54:25 +0100
> From: Artur Malabarba <bruce.connor.am@gmail.com>
> Cc: 20600@debbugs.gnu.org
> 
> Still, it sounds like it might be useful for run-hooks to properly handle
> functions whose arity is less than the expected arity.
> Does anyone foresee any issues with that?

To tell you the truth, I don't see the problem.  Is the problem the
need to specify a couple of arguments that will be left unused when
the function is called?  If so, we do it all over the place; we even
have a special convention for that.

bug#20600: 24.5.3: add-hook - wrong arguments error

[Artur Malabarba]

> To tell you the truth, I don't see the problem.  Is the problem the
> need to specify a couple of arguments that will be left unused when
> the function is called?  If so, we do it all over the place; we even
> have a special convention for that.

I don't see a problem either, I'm just suggesting an improvement on run-hooks.
The improvement would reduce the number of times you have to define a
wrapper around a function just to use it in a hook, by just not
passing to the function more arguments than it accepts.

Although, to be honest, I don't know how often that would be useful.
Even in the case originally mentioned above it would actually *not*
useful, because `save-buffer` *does* take an optional argument so the
wrapper around it would still be necessary.

bug#20600: 24.5.3: add-hook - wrong arguments error

[Stefan Monnier]

> The improvement would reduce the number of times you have to define a
> wrapper around a function just to use it in a hook, by just not
> passing to the function more arguments than it accepts.

Sounds dangerous and rarely useful.


        Stefan

bug#20600: 24.5.3: add-hook - wrong arguments error

[Artur Malabarba]

> Sounds dangerous and rarely useful.

Ok thanks. That's what I wanted to know. :-)

bug#20600: 24.5.3: add-hook - wrong arguments error

[Stefan Monnier]

> Still, it sounds like it might be useful for run-hooks to properly handle
> functions whose arity is less than the expected arity.

I have no idea what kind of "properly handle" you're thinking of.


        Stefan

bug#20600: 24.5.3: add-hook - wrong arguments error

[Artur Malabarba]

2015-05-18 17:47 GMT+01:00 Stefan Monnier

>> Still, it sounds like it might be useful for run-hooks to properly handle
>> functions whose arity is less than the expected arity.
>
> I have no idea what kind of "properly handle" you're thinking of.

(add-hook '1-arity-functions 'some-zero-arity-function)
(run-hooks '1-arity-functions some-arg)

Here, it will throw an error because run-hooks will try to call
(some-zero-arity-function some-arg). That's all fine and well
documented, I'm just saying it may be more convenient if run-hooks
took care to check that some-zero-arity-function has zero arity and
invoked it without arguments.

bug#20600: 24.5.3: add-hook - wrong arguments error

[Andreas Röhler]

Am 18.05.2015 um 19:29 schrieb Artur Malabarba:
> 2015-05-18 17:47 GMT+01:00 Stefan Monnier
>

Thanks all! Seems indeed rather a matter of understanding than a bug.
Can live with it.

close #20600