Question reg. docstring of symbol-value

Question reg. docstring of symbol-value

[Arash Esbati]

Hi all,

me and my fellow AUCTeX maintainers stumbled over the docstring of
`symbol-value' which says:

,----[ C-h f symbol-value RET ]
| symbol-value is a built-in function in ‘C source code’.
| 
| (symbol-value SYMBOL)
| 
| Return SYMBOL’s value.  Error if that is void.
| Note that if ‘lexical-binding’ is in effect, this returns the
| global value outside of any lexical scope.
`----

How does the last sentence fit to these two examples (eval'ed in
*scratch*):

--8<---------------cut here---------------start------------->8---
(defvar foo 1)
(let ((foo foo))
  (setq foo (1+ foo))
  (symbol-value 'foo))
;; => 2


(setq bar 1)
(let ((bar bar))
  (setq bar (1+ bar))
  (symbol-value 'bar))
;; => 1
--8<---------------cut here---------------end--------------->8---

Is the intention to say "..., this returns the global value outside of
any lexical scope for symbols not marked as special" or something like
that?  Or am I missing something?

Best, Arash

Re: Question reg. docstring of symbol-value

[Robin Tarsiger]

Arash Esbati wrote:
> Is the intention to say "..., this returns the global value outside of
> any lexical scope for symbols not marked as special" or something like
> that?  Or am I missing something?
 From the documentation for defvar:

 > The ‘defvar’ form also declares the variable as "special", so that it
 > is always dynamically bound even if ‘lexical-binding’ is t.

Basically, special variables don't get lexically bound in the first place,
so the let doesn't introduce a lexical binding for that variable; it
introduces a dynamic one, and dynamic bindings affect symbol-value.

-RTT

Re: Question reg. docstring of symbol-value

[Tassilo Horn]

Robin Tarsiger <rtt@dasyatidae.com> writes:

>> Is the intention to say "..., this returns the global value outside of
>> any lexical scope for symbols not marked as special" or something like
>> that?  Or am I missing something?
>
> From the documentation for defvar:
>
>> The ‘defvar’ form also declares the variable as "special", so that it
>> is always dynamically bound even if ‘lexical-binding’ is t.
>
> Basically, special variables don't get lexically bound in the first
> place, so the let doesn't introduce a lexical binding for that
> variable; it introduces a dynamic one, and dynamic bindings affect
> symbol-value.

Right, that's what we've observed and which we consider as expected
outcome.  The only situation where we've observed a difference in just
taking the value of a global variable or using `symbol-value' on its
symbol is for globals which came into being in terms of plain `setq'
instead `defvar' (which is bad practice anyway).

Is that really the only corner case where `global-var' and
`(symbol-value 'global-var)' can differ?  If so, the docstring (and the
info manual) could be much more specific.  If not, it would make sense
to name those situations, and provide examples in the manual.

Bye,
Tassilo