Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

[Stefan Monnier]

>     * src/window.c (save_window_save): When setting the marker
>     insertion type of saved window points treat a buffer local
>     value of 'unbound' for 'window-point-insertion-type' like
>     nil (Bug#33871).

I think treating it like nil is technically incorrect: instead of nil it
should use the global value of the variable (like Fbuffer_local_value would).


        Stefan

Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

[martin rudalics]

 >>      * src/window.c (save_window_save): When setting the marker
 >>      insertion type of saved window points treat a buffer local
 >>      value of 'unbound' for 'window-point-insertion-type' like
 >>      nil (Bug#33871).
 >
 > I think treating it like nil is technically incorrect: instead of nil it
 > should use the global value of the variable (like Fbuffer_local_value would).

The only difference between buffer_local_value and Fbuffer_local_value
I see is that the latter signals an error when the value returned by
buffer_local_value is 'unbound'.  I cannot signal an error here so I
use nil instead.  I don't see anything else I could do here but if you
have any suggestions I will try them.

In general, I am completely lost as to what the canonical way to find
out whether a variable has a buffer local binding is: I suppose it's

Fassoc (variable, BVAR (buf, local_var_alist), Qnil);

from buffer_local_value.  And I suppose in Lisp I would have to check
for membership in 'buffer-local-variables'.  Or how could I check the
following part of the doc-string of 'buffer-local-value' separately:

"If VARIABLE does not have a buffer-local binding in BUFFER,"

martin

Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

[Stefan Monnier]

> The only difference between buffer_local_value and Fbuffer_local_value
> I see is that the latter signals an error when the value returned by
> buffer_local_value is 'unbound'.  I cannot signal an error here so I
> use nil instead.  I don't see anything else I could do here but if you
> have any suggestions I will try them.

I don't understand why this var would be unbound unless someone call
`makunbound` on it.  Any idea?


        Stefan

Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

[martin rudalics]

 > I don't understand why this var would be unbound unless someone call
 > `makunbound` on it.  Any idea?

I think so, meanwhile.  Have a look at the definition of
Qwindow_point_insertion_type.

martin

Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

[Stefan Monnier]

>> I don't understand why this var would be unbound unless someone call
>> `makunbound` on it.  Any idea?
> I think so, meanwhile.  Have a look at the definition of
> Qwindow_point_insertion_type.

Hmm... I can't find the source of the problem...
Please tell us,


        Stefan

Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

[martin rudalics]

 >> Have a look at the definition of
 >> Qwindow_point_insertion_type.

 > Hmm... I can't find the source of the problem...
 > Please tell us,

   DEFSYM (Qwindow_point_insertion_type, "window_point_insertion_type");
                                                ^     ^         ^
martin

Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

[Stefan Monnier]

>>> Have a look at the definition of
>>> Qwindow_point_insertion_type.
>> Hmm... I can't find the source of the problem...
>> Please tell us,
>   DEFSYM (Qwindow_point_insertion_type, "window_point_insertion_type");
>                                                ^     ^         ^

Duh!


        Stefan

Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

[martin rudalics]

[-- Attachment #1: Type: text/plain, Size: 87 bytes --]

 > Duh!

Eli - anything against installing the attached on the release branch?

martin

[-- Attachment #2: window-point-insertion-type.diff --]
[-- Type: text/plain, Size: 557 bytes --]

--- a/src/window.c
+++ b/src/window.c
@@ -7646,7 +7646,7 @@ Value is a list of the form (WIDTH COLUMNS VERTICAL-TYPE HEIGHT LINES
   DEFVAR_LISP ("window-point-insertion-type", Vwindow_point_insertion_type,
 	       doc: /* Type of marker to use for `window-point'.  */);
   Vwindow_point_insertion_type = Qnil;
-  DEFSYM (Qwindow_point_insertion_type, "window_point_insertion_type");
+  DEFSYM (Qwindow_point_insertion_type, "window-point-insertion-type");
 
   DEFVAR_LISP ("window-configuration-change-hook",
 	       Vwindow_configuration_change_hook,

Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

[Eli Zaretskii]

> Date: Fri, 04 Jan 2019 11:00:15 +0100
> From: martin rudalics <rudalics@gmx.at>
> CC: emacs-devel@gnu.org, "eliz@gnu.org" <eliz@gnu.org>
> 
> Eli - anything against installing the attached on the release branch?

No, of course not.

Thanks for catching this blunder.

Re: [Emacs-diffs] master d82e73f: Handle 'unbound' like nil for 'window-point-insertion-type' (Bug#33871)

[martin rudalics]

>> Eli - anything against installing the attached on the release branch?
>
> No, of course not.
>
> Thanks for catching this blunder.

Installed.

Thanks, martin