Re: /srv/bzr/emacs/trunk r108127: New function set-temporary-overlay-map and macros (defvar|setq)-local.

Re: /srv/bzr/emacs/trunk r108127: New function set-temporary-overlay-map and macros (defvar|setq)-local.

[Leo]

On 2012-05-05 09:47 +0800, Stefan Monnier wrote:
> ------------------------------------------------------------
> +** New macros `setq-local' and `defvar-local'.

I think setq-local is a bad name jarring with setq and setq-default.

How about implement it as a function like this?

(defun set-local (symbol newval)
  "Set buffer-local value of SYMBOL to NEWVAL."
  (set (make-local-variable symbol) newval))

setq-local can then be properly implemented to accept a list of
alternate [SYM VAL]s just like every setq*.

Leo

Re: /srv/bzr/emacs/trunk r108127: New function set-temporary-overlay-map and macros (defvar|setq)-local.

[Stefan Monnier]

>> ------------------------------------------------------------
>> +** New macros `setq-local' and `defvar-local'.

> I think setq-local is a bad name jarring with setq and setq-default.

Why "jarring"?  I actually used setq-local, specifically because it is
similar to setq-default.

> How about implement it as a function like this?

> (defun set-local (symbol newval)
>   "Set buffer-local value of SYMBOL to NEWVAL."
>   (set (make-local-variable symbol) newval))

I've never seen this construction used with anything but a static
variable, so I don't think the difference matters much,


        Stefan

Re: /srv/bzr/emacs/trunk r108127: New function set-temporary-overlay-map and macros (defvar|setq)-local.

[Johan Bockgård]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>> ------------------------------------------------------------
>>> +** New macros `setq-local' and `defvar-local'.
>
>> I think setq-local is a bad name jarring with setq and setq-default.
>
> Why "jarring"?  I actually used setq-local, specifically because it is
> similar to setq-default.

FWIW, I do find `set-temporary-overlay-map'--which has nothing to do
with overlays--jarring.

Re: /srv/bzr/emacs/trunk r108127: New function set-temporary-overlay-map and macros (defvar|setq)-local.

[Leo]

On 2012-05-05 11:37 +0800, Stefan Monnier wrote:
> Why "jarring"?  I actually used setq-local, specifically because it is
> similar to setq-default.

The fact that (setq-local var1 val1 var2 val2 ...) doesn't work.

Leo

Re: /srv/bzr/emacs/trunk r108127: New function set-temporary-overlay-map and macros (defvar|setq)-local.

[Stefan Monnier]

>> Why "jarring"?  I actually used setq-local, specifically because it is
>> similar to setq-default.
> The fact that (setq-local var1 val1 var2 val2 ...) doesn't work.

Oh, yes, I always disliked this use, so obviously I didn't add it to
setq-local.


        Stefan

Re: /srv/bzr/emacs/trunk r108127: New function set-temporary-overlay-map and macros (defvar|setq)-local.

[Leo]

On 2012-05-06 09:39 +0800, Stefan Monnier wrote:
> Oh, yes, I always disliked this use, so obviously I didn't add it to
> setq-local.

I don't like that much either. but I am concerned about consistency
between these setq* special forms.

Leo

Re: /srv/bzr/emacs/trunk r108127: New function set-temporary-overlay-map and macros (defvar|setq)-local.

[Leo]

On 2012-05-05 21:05 +0800, Johan Bockgård wrote:
> FWIW, I do find `set-temporary-overlay-map'--which has nothing to do
> with overlays--jarring.

Stefan, did you mean "overriding"?

Leo

Re: /srv/bzr/emacs/trunk r108127: New function set-temporary-overlay-map and macros (defvar|setq)-local.

[Stefan Monnier]

>> FWIW, I do find `set-temporary-overlay-map'--which has nothing to do
>> with overlays--jarring.
> Stefan, did you mean "overriding"?

Contrary to overriding-local-map and overriding-terminal-local-map,
set-temporary-overlay-map installs a map which does not disable
other maps (i.e. as soon as overriding-local-map is non-nil, most maps
(other than the global-map) become inactive, but not so with
set-temporary-overlay-map).


        Stefan

Re: /srv/bzr/emacs/trunk r108127: New function set-temporary-overlay-map and macros (defvar|setq)-local.

[Nix]

On 6 May 2012, Leo told this:

> On 2012-05-06 09:39 +0800, Stefan Monnier wrote:
>> Oh, yes, I always disliked this use, so obviously I didn't add it to
>> setq-local.
>
> I don't like that much either. but I am concerned about consistency
> between these setq* special forms.

More generally, some people (e.g. me) *do* like it, and are unlikely to
use setq-local in our own code if it doesn't support something similar.

-- 
NULL && (void)

RE: /srv/bzr/emacs/trunk r108127: New functionset-temporary-overlay-map and macros (defvar|setq)-local.

[Drew Adams]

> >> Oh, yes, I always disliked this use, so obviously I didn't 
> >> add it to setq-local.
> >
> > I don't like that much either. but I am concerned about consistency
> > between these setq* special forms.
> 
> More generally, some people (e.g. me) *do* like it, and are 
> unlikely to use setq-local in our own code if it doesn't support 
> something similar.

Not that I really care much about this... but FWIW -

For years I tended not to use `setq' for more than one assignment, feeling much
as what Stefan apparently feels.  A few years back I started to change (I don't
recall why).  And now I'm pretty much on the other side of the fence.

In general, I now find the multiple-assignment form more readable (less noise)
and lower maintenance.  It is especially more readable to group assignments if
they are closely related.  But even just in terms of general readability, I find
A better than B:

A:

(if whatever
    (setq alpha       (+ (beta) gamma)
          delta       epsilon
          iota-omega  (pool-fni dne-on))
  (do-this)
  (do-that)
  (do-more)
  (enough-already))

B:

(if whatever
    (progn
     (setq alpha       (+ (beta) gamma))
     (setq delta       epsilon
     (setq iota-omega  (pool-fni dne-on)))
  (do-this)
  (do-that)
  (do-more)
  (enough-already))

Again, FWIW/YMMV.  And if you argue that without clear layout it can be
confusing, I agree.  Which is why I don't write it this way:

(setq alpha (+ (beta) gamma) delta
      epsilon iota-omega (pool-fni dne-on))

So I think it's good for Emacs to support such syntax for `setq-local', other
things being equal (are they?).  Nothing requires anyone to use it.  And it is
consistent with setq, etc.