set-window-dedicated-p argument name

set-window-dedicated-p argument name

[Juanma Barranquero]

Martin, did you know about this old thread

http://lists.gnu.org/archive/html/emacs-devel/2007-05/msg01358.html

before doing the following change?

2008-10-18  Martin Rudalics  <rudalics@gmx.at>

	* window.c [...]
	(Fwindow_dedicated_p): Make window argument optional.
	(Fset_window_dedicated_p): Rename argument "arg" to "flag".

  Juanma

Re: set-window-dedicated-p argument name

[martin rudalics]

 > Martin, did you know about this old thread
 >
 > http://lists.gnu.org/archive/html/emacs-devel/2007-05/msg01358.html
 >
 > before doing the following change?
 >
 > 2008-10-18  Martin Rudalics  <rudalics@gmx.at>
 >
 > 	* window.c [...]
 > 	(Fwindow_dedicated_p): Make window argument optional.
 > 	(Fset_window_dedicated_p): Rename argument "arg" to "flag".

I didn't know about that thread.  AFAICT I have not broken anything here
with respect to how this is documented.  If we want to consider a window
dedicated iff `window-dedicated-p' equals t, we have to explicitly say
that in the doc-string and the manual and then we can (re-)implement
this or anything else that suits Stefan's needs.  Until now, it was
possible to do

(set-window-dedicated-p (selected-window) 1)

and the window was _not_ considered dedicated by some functions.  That
was in clear contradiction to what the documentation said.

martin

Re: set-window-dedicated-p argument name

[Juanma Barranquero]

On Wed, Oct 22, 2008 at 11:35, martin rudalics <rudalics@gmx.at> wrote:

> I didn't know about that thread.  AFAICT I have not broken anything here
> with respect to how this is documented.

I don't think the previous docstrings were good
(set-window-dedicated-p talked about a "dedication flag", while
window-dedicated-p referred to a "dedicated object", which was nowhere
explained). That said, with your wording it is not clear now that
w-d-p returns the object passed to s-w-d-p.

> If we want to consider a window
> dedicated iff `window-dedicated-p' equals t, we have to explicitly say
> that in the doc-string and the manual and then we can (re-)implement
> this or anything else that suits Stefan's needs.

I think at the moment the thing to do is consider a window dedicated
whenever FLAG is non-nil. If and when Stefan adds "different levels of
dedication", code affected by window dedication should be revised.

> Until now, it was
> possible to do
>
> (set-window-dedicated-p (selected-window) 1)
>
> and the window was _not_ considered dedicated by some functions.

I'd say that is a bug of these functions.

  Juanma

Re: set-window-dedicated-p argument name

[martin rudalics]

 > I don't think the previous docstrings were good
 > (set-window-dedicated-p talked about a "dedication flag", while
 > window-dedicated-p referred to a "dedicated object", which was nowhere
 > explained). That said, with your wording it is not clear now that
 > w-d-p returns the object passed to s-w-d-p.

You're right.  I merely adapted the nomenclature of the Elisp manual
which did not talk about a "dedicated object".

 >> If we want to consider a window
 >> dedicated iff `window-dedicated-p' equals t, we have to explicitly say
 >> that in the doc-string and the manual and then we can (re-)implement
 >> this or anything else that suits Stefan's needs.
 >
 > I think at the moment the thing to do is consider a window dedicated
 > whenever FLAG is non-nil.

That's how I did interpret it.

 > If and when Stefan adds "different levels of
 > dedication", code affected by window dedication should be revised.

I agree that different levels of dedication might make sense.  In
particular, `set-window-buffer' could reuse a dedicated window if
`window-dedicated-p' is something "weaker" than t.

 >> Until now, it was
 >> possible to do
 >>
 >> (set-window-dedicated-p (selected-window) 1)
 >>
 >> and the window was _not_ considered dedicated by some functions.
 >
 > I'd say that is a bug of these functions.

Which I hopefully corrected with my patch.

martin

Re: set-window-dedicated-p argument name

[Stefan Monnier]

> I think at the moment the thing to do is consider a window dedicated
> whenever FLAG is non-nil. If and when Stefan adds "different levels of
> dedication", code affected by window dedication should be revised.

Actually, the distinction is there already.  What's missing is
documentation for it.

>> Until now, it was
>> possible to do
>> 
>> (set-window-dedicated-p (selected-window) 1)
>> 
>> and the window was _not_ considered dedicated by some functions.

> I'd say that is a bug of these functions.

No, it's a feature.


        Stefan

Re: set-window-dedicated-p argument name

[Juanma Barranquero]

On Wed, Oct 22, 2008 at 15:34, Stefan Monnier <monnier@iro.umontreal.ca> wrote:

>> I'd say that is a bug of these functions.
>
> No, it's a feature.

Ah, yes, of course it is a feature if the distinction is there already.

  Juanma