default-truncate-lines (was: Changes to emacs/lisp/gnus/gnus-art.el, v)

default-truncate-lines (was: Changes to emacs/lisp/gnus/gnus-art.el, v)

[Reiner Steib]

On Thu, Nov 22 2007, Stefan Monnier changed:

> --- gnus-art.el	16 Nov 2007 16:50:34 -0000	1.139
> +++ gnus-art.el	22 Nov 2007 20:13:35 -0000	1.140
[...]
> -(defcustom gnus-article-truncate-lines default-truncate-lines
> +(defcustom gnus-article-truncate-lines (default-value 'truncate-lines)
>    "Value of `truncate-lines' in Gnus Article buffer.

I don't care about this change, but out of curiosity: (why) does it
make a difference?

,----[ <f1> v default-truncate-lines RET ]
| default-truncate-lines is a variable defined in `src/buffer.c'.
| Its value is nil
| 
| Documentation:
| Default value of `truncate-lines' for buffers that do not override it.
| This is the same as (default-value 'truncate-lines).
`----

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: default-truncate-lines

[Stefan Monnier]

>> -(defcustom gnus-article-truncate-lines default-truncate-lines
>> +(defcustom gnus-article-truncate-lines (default-value 'truncate-lines)
> I don't care about this change, but out of curiosity: (why) does it
> make a difference?

Every time I see one of those default-FOO variable used I get flashes of
the following code:

	  /* If this variable is a default for something stored
	     in the buffer itself, such as default-fill-column,
	     find the buffers that don't have local values for it
	     and update them.  */
	  if (XOBJFWD (valcontents)->objvar > (Lisp_Object *) &buffer_defaults
	      && XOBJFWD (valcontents)->objvar < (Lisp_Object *) (&buffer_defaults + 1))
	    {
	      int offset = ((char *) XOBJFWD (valcontents)->objvar
			    - (char *) &buffer_defaults);
	      int idx = PER_BUFFER_IDX (offset);

	      Lisp_Object tail;

	      if (idx <= 0)
		break;

	      for (tail = Vbuffer_alist; CONSP (tail); tail = XCDR (tail))
		{
		  Lisp_Object buf;
		  struct buffer *b;

		  buf = Fcdr (XCAR (tail));
		  if (!BUFFERP (buf)) continue;
		  b = XBUFFER (buf);

		  if (! PER_BUFFER_VALUE_P (b, idx))
		    PER_BUFFER_VALUE (b, offset) = newval;
		}
	    }

and every time I get to replace it with a use of `default-value', I feel
like maybe this madness will disappear some day.


        Stefan

Re: default-truncate-lines

[Richard Stallman]

	      /* If this variable is a default for something stored
		 in the buffer itself, such as default-fill-column,
		 find the buffers that don't have local values for it
		 and update them.  */

The reason for this is so that variables like `fill-column'
which are stored in slots in the buffer
can have buffer-local values in some buffers and not in others.
Setting the default value needs to store it into the proper
slot in buffers that don't have buffer-local bindings of the variable.

The code seems straightforward.  I don't see how use of
`default-value' could replace it.

Re: default-truncate-lines

[Stefan Monnier]

> 	      /* If this variable is a default for something stored
> 		 in the buffer itself, such as default-fill-column,
> 		 find the buffers that don't have local values for it
> 		 and update them.  */

> The reason for this is so that variables like `fill-column'
> which are stored in slots in the buffer
> can have buffer-local values in some buffers and not in others.
> Setting the default value needs to store it into the proper
> slot in buffers that don't have buffer-local bindings of the variable.

> The code seems straightforward.  I don't see how use of
> `default-value' could replace it.

Yes, I've spent enough time recently looking at the var manipulation
code to know that it's correct, but this whole code is only there for
the default-FOO variables.  The loop over all buffers is also present in
set-default where it's indeed necessary.  We could/should of course
use a separate function to avoid this redundancy, but the main ugliness
is in the first 5 lines (which are not needed in set-default):

	  if (XOBJFWD (valcontents)->objvar > (Lisp_Object *) &buffer_defaults
	      && XOBJFWD (valcontents)->objvar < (Lisp_Object *) (&buffer_defaults + 1))
	    {
	      int offset = ((char *) XOBJFWD (valcontents)->objvar
			    - (char *) &buffer_defaults);

comparing pointers like that is terribly hackish.


        Stefan

Re: default-truncate-lines

[Reiner Steib]

On Fri, Nov 23 2007, Stefan Monnier wrote:

>>> -(defcustom gnus-article-truncate-lines default-truncate-lines
>>> +(defcustom gnus-article-truncate-lines (default-value 'truncate-lines)
>> I don't care about this change, but out of curiosity: (why) does it
>> make a difference?
>
> Every time I see one of those default-FOO variable used I get flashes of
> the following code:
[...]
> and every time I get to replace it with a use of `default-value', I feel
> like maybe this madness will disappear some day.

Wouldn't it make sense to suggest using (default-value 'foo) in the
doc strings of `default-foo'?

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: default-truncate-lines

[Stefan Monnier]

>>>> -(defcustom gnus-article-truncate-lines default-truncate-lines
>>>> +(defcustom gnus-article-truncate-lines (default-value 'truncate-lines)
>>> I don't care about this change, but out of curiosity: (why) does it
>>> make a difference?
>> 
>> Every time I see one of those default-FOO variable used I get flashes of
>> the following code:
> [...]
>> and every time I get to replace it with a use of `default-value', I feel
>> like maybe this madness will disappear some day.

> Wouldn't it make sense to suggest using (default-value 'foo) in the
> doc strings of `default-foo'?

Indeed.  We should just mark all those default-* variables obsolete.
The only problematic one, AFAICT is default-enable-multibyte-characters
since (setq-default 'enable-multibyte-characters) doesn't work.


        Stefan

Re: default-truncate-lines

[Richard Stallman]

    Indeed.  We should just mark all those default-* variables obsolete.

Why do that?  Some people may use them; is there any positive benefit
that justifies making them change?