The two-argument form of defvar

The two-argument form of defvar

[Philipp Stephani]

Hi,

when looking at the source code of defvar it becomes clear that the
two-argument form

(defvar foo)

is a no-op. It is often used as a signal to the byte compiler that a
certain variable will be available, similar to declare-function. Would it
be possible to clarify the docstring accordingly? Currently it has no
indication that the two-argument form doesn't do anything.

Re: The two-argument form of defvar

[Nicolas Richard]

Philipp Stephani <p.stephani2@gmail.com> writes:

> Hi,
>
> when looking at the source code of defvar it becomes clear that the
> two-argument form
>
> (defvar foo)
>
> is a no-op.

Not always a no-op. The source code has this comment :
    /* A simple (defvar foo) with lexical scoping does "nothing" except
       declare that var to be dynamically scoped *locally* (i.e. within
       the current file or let-block).  */

To reflect the above comment, the docstring has :
| The `defvar' form also declares the variable as "special",
| so that it is always dynamically bound even if `lexical-binding' is t.

-- 
Nicolas Richard

Re: The two-argument form of defvar

[Tassilo Horn]

Nicolas Richard <theonewiththeevillook@yahoo.fr> writes:

>> when looking at the source code of defvar it becomes clear that the
>> two-argument form
>>
>> (defvar foo)
>>
>> is a no-op.
>
> Not always a no-op. The source code has this comment :
>     /* A simple (defvar foo) with lexical scoping does "nothing" except
>        declare that var to be dynamically scoped *locally* (i.e. within
>        the current file or let-block).  */
>
> To reflect the above comment, the docstring has :
> | The `defvar' form also declares the variable as "special",
> | so that it is always dynamically bound even if `lexical-binding' is t.

But it doesn't *define* anything.

  (defvar xxx1 1) ;; C-h v xxx1 works
  (defvar xxx2)   ;; C-h v xxx2<tab> (No matches)

That behavior is a bit unexpected when the docstring says "Define SYMBOL
as a variable, and return SYMBOL."

Bye,
Tassilo

Re: The two-argument form of defvar

[Philipp Stephani]

Maybe there should be a declare-variable function to forward-declare
variables, parallel to declare-function.

Tassilo Horn <tsdh@gnu.org> schrieb am Do., 19. März 2015 um 16:52 Uhr:

> Nicolas Richard <theonewiththeevillook@yahoo.fr> writes:
>
> >> when looking at the source code of defvar it becomes clear that the
> >> two-argument form
> >>
> >> (defvar foo)
> >>
> >> is a no-op.
> >
> > Not always a no-op. The source code has this comment :
> >     /* A simple (defvar foo) with lexical scoping does "nothing" except
> >        declare that var to be dynamically scoped *locally* (i.e. within
> >        the current file or let-block).  */
> >
> > To reflect the above comment, the docstring has :
> > | The `defvar' form also declares the variable as "special",
> > | so that it is always dynamically bound even if `lexical-binding' is t.
>
> But it doesn't *define* anything.
>
>   (defvar xxx1 1) ;; C-h v xxx1 works
>   (defvar xxx2)   ;; C-h v xxx2<tab> (No matches)
>
> That behavior is a bit unexpected when the docstring says "Define SYMBOL
> as a variable, and return SYMBOL."
>
> Bye,
> Tassilo
>

Re: The two-argument form of defvar

[Stefan Monnier]

>> (defvar xxx2)   ;; C-h v xxx2<tab> (No matches)

What kind of info would you expect to get from C-h v xxx2 RET?
All Emacs knows about it at that point is that you intend to use it.
So C-h v wouldn't be able to tell you anything interesting about it.


        Stefan

RE: The two-argument form of defvar

[Drew Adams]

(FWIW, this thread is about the one-argument form of defvar,
not a two-argument form.)

Re: The two-argument form of defvar

[Philipp Stephani]

Drew Adams <drew.adams@oracle.com> schrieb am Sa., 18. Apr. 2015 um
16:14 Uhr:

> (FWIW, this thread is about the one-argument form of defvar,
> not a two-argument form.)
>

I intended it to be about the one-argument form, but got the subject wrong.
Probably a habit from other programming languages where the number of
identifiers between ( and ) is the number of arguments.

Re: The two-argument form of defvar

[Emanuel Berg]

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

>>> (defvar xxx2) ;; C-h v xxx2<tab> (No matches)
>
> What kind of info would you expect to get from C-h
> v xxx2 RET? All Emacs knows about it at that point
> is that you intend to use it. So C-h v wouldn't be
> able to tell you anything interesting about it.

It could say precisely that - no value, not documented,
only the intention: to use it. It doesn't, and I don't
suggest it should, but it isn't an obviously wrong
thing to do.

-- 
underground experts united
http://user.it.uu.se/~embe8573