buffer local part of a hook

buffer local part of a hook

[Matthias Pfeifer]

Hi there,

i stumbled over this pieace of information in the "gnu emacs lisp reference
manual":

"The buffer-local part of this hook is run once for each window on the
affected frame, with the relevant window selected and its buffer current.
The global part is run once for the modified frame, with that frame
selected."

Question is: What is the buffer local / global part of a hook?



matthias.

Re: buffer local part of a hook

[Dmitry Gutov]

On 03/07/2016 02:52 PM, Matthias Pfeifer wrote:

> Question is: What is the buffer local / global part of a hook?

The buffer local and global values of the hook variable. Although that 
should probably be considered an implementation detail.

Re: buffer local part of a hook

[Stefan Monnier]

>> Question is: What is the buffer local / global part of a hook?
> The buffer local and global values of the hook variable. Although that
> should probably be considered an implementation detail.

No, it's part of the exposed functionality, since add/remove-hook
specify explicitly with the `local' argument whether the specified
function should be added-to/removed-from the global or the local part.


        Stefan

Re: buffer local part of a hook

[Dmitry Gutov]

On 03/07/2016 04:36 PM, Stefan Monnier wrote:

> No, it's part of the exposed functionality, since add/remove-hook
> specify explicitly with the `local' argument whether the specified
> function should be added-to/removed-from the global or the local part.

So we're fine with users changing the values of hooks directly, without 
using add-hook/remove-hook?

Re: buffer local part of a hook

[Stefan Monnier]

>> No, it's part of the exposed functionality, since add/remove-hook
>> specify explicitly with the `local' argument whether the specified
>> function should be added-to/removed-from the global or the local part.
> So we're fine with users changing the values of hooks directly, without
> using add-hook/remove-hook?

No.  But we're fine with exposing the concept that a hook as a "local
part" and a "global part".


        Stefan

Re: buffer local part of a hook

[Dmitry Gutov]

On 03/07/2016 07:31 PM, Stefan Monnier wrote:

> No.  But we're fine with exposing the concept that a hook as a "local
> part" and a "global part".

The original question, how I understood it, was how both parts are stored.

Re: buffer local part of a hook

[Stefan Monnier]

>> No.  But we're fine with exposing the concept that a hook as a "local
>> part" and a "global part".
> The original question, how I understood it, was how both parts are stored.

I understood it to just asking for an explanation of those concepts are,
rather than how they're implemented.  As for "exposing the
implementation", while I don't really like it we do already document it
in the Elisp manual:

 -- Function: run-hooks &rest hookvars
     [...]
     If the hook variable is buffer-local, the buffer-local variable
     will be used instead of the global variable.  However, if the
     buffer-local variable contains the element ‘t’, the global hook
     variable will be run as well.


-- Stefan

Re: buffer local part of a hook

[Dmitry Gutov]

On 03/08/2016 04:49 PM, Stefan Monnier wrote:

>> The original question, how I understood it, was how both parts are stored.
>
> I understood it to just asking for an explanation of those concepts are,
> rather than how they're implemented.

In that case, we still haven't explained them, I think (and I'm not sure 
which words to use).

> As for "exposing the
> implementation", while I don't really like it we do already document it
> in the Elisp manual:

That's unfortunate. Has anything comparable, ever, been un-documented, 
in the history of Emacs?

Re: buffer local part of a hook

[Stefan Monnier]

> In that case, we still haven't explained them, I think (and I'm not sure
> which words to use).

Could be.

>> As for "exposing the implementation", while I don't really like it we
>> do already document it in the Elisp manual:
> That's unfortunate.  Has anything comparable, ever, been un-documented, in
> the history of Emacs?

No idea.  But s long as we don't intend to change the implementation,
it's not really a problem,


        Stefan