Re: [Emacs-diffs] master 8b36bfa 2/2: Add autoload cookie for the eww-suggest-uris variable.

Re: [Emacs-diffs] master 8b36bfa 2/2: Add autoload cookie for the eww-suggest-uris variable.

[Artur Malabarba]

> +;;;###autoload
>  (defcustom eww-suggest-uris

Autoloading defcustoms is know to cause problems. See
http://lists.gnu.org/archive/html/emacs-devel/2014-11/msg00608.html

Personally, I'd recommend finding another way around your add-hook
issue. What was the problem?

autoload for eww-suggest-uris

[Ivan Shmakov]

>>>>> Artur Malabarba <bruce.connor.am@gmail.com> writes:

 >> +;;;###autoload
 >>  (defcustom eww-suggest-uris

 > Autoloading defcustoms is know to cause problems.  See
 > http://lists.gnu.org/archive/html/emacs-devel/2014-11/msg00608.html

	I don’t seem to see there any description of what /exactly/
	happens in such a case – just that package-initialize for some
	reason resets the variables set with custom-set-variables.

	Now, given that eww.el is a built-in package, is that even a
	problem?  (Any details on that, anyone?)

 > Personally, I'd recommend finding another way around your add-hook
 > issue.  What was the problem?

	This:

;;; ~/.emacs
(add-hook 'eww-suggest-uris 'some-new-function)
; → (some-new-function)

	Expected: that some-new-function /and/ all the default ones.

	An alternative would be to use add-hook in 'eww itself, but that
	in turn interferes with (setq eww-suggest-uris '(the-only-one)).

-- 
FSF associate member #7257  http://boycottsystemd.org/  … 3013 B6A0 230E 334A

Re: autoload for eww-suggest-uris

[Artur Malabarba]

>  > Personally, I'd recommend finding another way around your add-hook
>  > issue.  What was the problem?
>
>         This:
>
> ;;; ~/.emacs
> (add-hook 'eww-suggest-uris 'some-new-function)
> ; → (some-new-function)
>
>         Expected: that some-new-function /and/ all the default ones.
>
>         An alternative would be to use add-hook in 'eww itself, but that
>         in turn interferes with (setq eww-suggest-uris '(the-only-one)).

According to Stefan it is a bug to define a hook with a non-nil
initial value (see
http://lists.gnu.org/archive/html/emacs-devel/2015-02/msg01271.html),
so using the `add-hook' in eww itself is the correct way.
In fact, it seems that 25.1 will issue a compiler warning when a hook
variable is given a non-nil initial value.

Re: autoload for eww-suggest-uris

[Ivan Shmakov]

>>>>> Artur Malabarba <bruce.connor.am@gmail.com> writes:
>>>>> "SM" == Stefan Monnier <monnier@iro.umontreal.ca> writes:

[…]

 >> ;;; ~/.emacs (add-hook 'eww-suggest-uris 'some-new-function)
 >> ; → (some-new-function)

 >> Expected: that some-new-function /and/ all the default ones.

 >> An alternative would be to use add-hook in 'eww itself, but that in
 >> turn interferes with (setq eww-suggest-uris '(the-only-one)).

 > According to Stefan it is a bug to define a hook with a non-nil
 > initial value (see
 > http://lists.gnu.org/archive/html/emacs-devel/2015-02/msg01271.html),

	[AKA news:jwvd250fej3.fsf-monnier+emacs@gnu.org.]

 > so using the `add-hook' in eww itself is the correct way.

	How do we prevent it from interfering with an explicit ‘setq’
	then?

 SM> Basically, given that "unbound == nil", what is a bug is to defvar
 SM> a hook with a non-nil default value, unless that hook is predefined
 SM> (i. e. is never unbound).

	How exactly does this affect the case when autoloads /are/
	properly set up (as, I argue, they should be, assuming a proper
	Emacs installation)?

 > In fact, it seems that 25.1 will issue a compiler warning when a hook
 > variable is given a non-nil initial value.

	Rest assured I’ll promptly ignore it whenever working on my own
	code.

-- 
FSF associate member #7257  http://boycottsystemd.org/  … 3013 B6A0 230E 334A

Re: autoload for eww-suggest-uris

[Stefan Monnier]

> 	How do we prevent it from interfering with an explicit ‘setq’
> 	then?

The question is fundamentally: is eww-suggest-uris a hook, or is it
a list of functions?

If it's a hook, then using `setq' on it is a bug (and it should be
called eww-suggest-uris-hook or eww-suggest-uris-functions).


        Stefan

Re: autoload for eww-suggest-uris

[Ivan Shmakov]

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

 >> How do we prevent it from interfering with an explicit ‘setq’ then?

 > The question is fundamentally: is eww-suggest-uris a hook, or is it a
 > list of functions?

 > If it's a hook, then using `setq' on it is a bug (and it should be
 > called eww-suggest-uris-hook or eww-suggest-uris-functions).

	If we define a hook as a variable holding a list of functions
	and provided with the purpose of allowing proper operation of
	arbitrary extensions to the base code, then no – I believe
	eww-suggest-uris is /not/ a hook.  Still, it’s a (customizable)
	list of functions, defined with :type 'hook, and (arguably) one
	may use add-hook in one’s ~/.emacs to add to it.  (The use of
	add-hook against this variable in EWW extensions – should there
	be any – is debatable, though.)

-- 
FSF associate member #7257  np. La Druido — Strika Tango  3013 B6A0 230E 334A