globalized minor modes - priority over mode hook?

globalized minor modes - priority over mode hook?

[David Reitter]

`run-mode-hooks' runs the mode hooks first, then after-change-major-mode-hooks.

When a globalized major mode is enabled, then it will install itself in `after-change-major-mode-hooks' and override the more specific major mode hooks, which is not normally what one wants.

For instance, if I want global-visual-line-mode on, but turn off word-wrap in my `latex-mode-hook' in favor for auto-fill, I don't get the expected result.  I don't know what happens when local variables are set when a file is loaded - these are specific for the file and should probably override the less-specific global minor mode.

Could globalized minor modes not be enabled earlier?

Re: globalized minor modes - priority over mode hook?

[Stefan Monnier]

> `run-mode-hooks' runs the mode hooks first, then
> after-change-major-mode-hooks.

That would be a bug.  Do you have a recipe to reproduce it?


        Stefan

Re: globalized minor modes - priority over mode hook?

[David Reitter]

On Apr 26, 2010, at 10:30 AM, Stefan Monnier wrote:

>> `run-mode-hooks' runs the mode hooks first, then
>> after-change-major-mode-hooks.
> 
> That would be a bug.  Do you have a recipe to reproduce it?

Yes, below.

Looking at the code of `after-find-file', I think file-local variables will similarly be overruled by the global minor mode.

I ended up implementing what I wanted below by setting defaults for `auto-fill-function', but that's more low-level than intended, obviously.

I'm bcc'ing the bug DB.

---
Emacs -Q

(add-hook 'text-mode-hook 'turn-on-word-wrap) ; turns OFF auto-fill

(defvaralias 'auto-fill-mode 'auto-fill-function)  ; kludge
(define-globalized-minor-mode global-auto-fill-mode auto-fill-mode turn-on-auto-fill)

(defun turn-on-word-wrap ()
  "Turn on Word Wrap mode in current buffer."
  (turn-off-auto-fill)
  (turn-on-visual-line-mode))

;; demo

(global-auto-fill-mode 1)  ; set default
(text-mode)

;; result: auto-fill-mode is on, even though it is intended to be off

Re: globalized minor modes - priority over mode hook?

[Glenn Morris]

David Reitter wrote:

> I'm bcc'ing the bug DB.

BCC'ing the bug tracker doesn't work for creating new reports.
It has no idea what package to associate your report with, since it
relies on the To:/Cc: header to figure this out.
Hence 6040 got assigned to debbugs.gnu.org (the default) rather than
emacs.

Re: globalized minor modes - priority over mode hook?

[Stefan Monnier]

>>> `run-mode-hooks' runs the mode hooks first, then
>>> after-change-major-mode-hooks.
>> That would be a bug.  Do you have a recipe to reproduce it?
> Yes, below.

Oops, sorry, I misunderstood.
Yes, indeed, it will be run last.  And it will also be run first if the
mode is written "properly" (by inheriting from some other mode).
We should maybe split this into two parts: fundamental-mode-hook and
after-change-major-mode-hook, the first is the one run before any other
mode hook and the second is run after all other mode hooks.


        Stefan

Re: globalized minor modes - priority over mode hook?

[Lennart Borgman]

On Mon, Apr 26, 2010 at 7:56 PM, Stefan Monnier
<monnier@iro.umontreal.ca> wrote:
>>>> `run-mode-hooks' runs the mode hooks first, then
>>>> after-change-major-mode-hooks.
>>> That would be a bug.  Do you have a recipe to reproduce it?
>> Yes, below.
>
> Oops, sorry, I misunderstood.
> Yes, indeed, it will be run last.  And it will also be run first if the
> mode is written "properly" (by inheriting from some other mode).
> We should maybe split this into two parts: fundamental-mode-hook and
> after-change-major-mode-hook, the first is the one run before any other
> mode hook and the second is run after all other mode hooks.

Sounds great.

Re: globalized minor modes - priority over mode hook?

[Juanma Barranquero]

On Mon, Apr 26, 2010 at 19:56, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
>
> Oops, sorry, I misunderstood.
> Yes, indeed, it will be run last.  And it will also be run first if the
> mode is written "properly" (by inheriting from some other mode).
> We should maybe split this into two parts: fundamental-mode-hook and
> after-change-major-mode-hook, the first is the one run before any other
> mode hook and the second is run after all other mode hooks.

Has this been fixed by r100061?

Because, for example,

  (add-hook 'completion-list-mode-hook (lambda () (linum-mode -1)))
  (global-linum-mode 1)

still activates linum-mode in completion buffers.

    Juanma

Re: globalized minor modes - priority over mode hook?

[Stefan Monnier]

>> Oops, sorry, I misunderstood.
>> Yes, indeed, it will be run last.  And it will also be run first if the
>> mode is written "properly" (by inheriting from some other mode).
>> We should maybe split this into two parts: fundamental-mode-hook and
>> after-change-major-mode-hook, the first is the one run before any other
>> mode hook and the second is run after all other mode hooks.
> Has this been fixed by r100061?

Not yet, no.


        Stefan

Re: globalized minor modes - priority over mode hook?

[Stefan Monnier]

> Because, for example,

>   (add-hook 'completion-list-mode-hook (lambda () (linum-mode -1)))
>   (global-linum-mode 1)

> still activates linum-mode in completion buffers.

I believe my last changes do make it work right in this case.


        Stefan

Re: globalized minor modes - priority over mode hook?

[Juanma Barranquero]

On Wed, Apr 28, 2010 at 23:53, Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> I believe my last changes do make it work right in this case.

Yes, they do.

As an aside, it would be good to have a simple way to say "I don't
want global mode X applied to buffers Y, Z and W." Otherwise,
disabling and reenabling the global-minor-mode turns it on again on
buffers that previously disabled it via the MODE-hook.

But that's a different problem that the one you just fixed.

Thanks,

    Juanma

Re: globalized minor modes - priority over mode hook?

[Stefan Monnier]

>> I believe my last changes do make it work right in this case.
> Yes, they do.

Thanks for confirming.

> As an aside, it would be good to have a simple way to say "I don't
> want global mode X applied to buffers Y, Z and W."

Yes, that would be a welcome addition.


        Stefan

Re: globalized minor modes - priority over mode hook?

[David Reitter]

On Apr 28, 2010, at 6:39 PM, Stefan Monnier wrote:
>> As an aside, it would be good to have a simple way to say "I don't
>> want global mode X applied to buffers Y, Z and W."
> 
> Yes, that would be a welcome addition.

I agree.  

Please consider hand-crafting global minor modes in some cases.

Take `visual-line-mode': What it does is to configure a number of variables, most of which are buffer-local:

(line-move-visual truncate-lines truncate-partial-width-windows word-wrap fringe-indicator-alist)

The classic way to configure a default for these would be `set-default'.  This seems cleaner and more efficient than automatically deriving a globalized version of the mode that runs (at least) at every major mode change and may then be turned off again by the mode hook. 

bug#6040: globalized minor modes - priority over mode hook?

[Glenn Morris]

Version: 24.1

David Reitter wrote:

> Emacs -Q
>
> (add-hook 'text-mode-hook 'turn-on-word-wrap) ; turns OFF auto-fill
>
> (defvaralias 'auto-fill-mode 'auto-fill-function)  ; kludge
> (define-globalized-minor-mode global-auto-fill-mode auto-fill-mode turn-on-auto-fill)
>
> (defun turn-on-word-wrap ()
>   "Turn on Word Wrap mode in current buffer."
>   (turn-off-auto-fill)
>   (turn-on-visual-line-mode))
>
> ;; demo
>
> (global-auto-fill-mode 1)  ; set default
> (text-mode)
>
> ;; result: auto-fill-mode is on, even though it is intended to be off


I see this in 23.3 but not the current trunk, so it looks like this was
fixed.