Re: Weird error message "wrong-type-argument symbolp (quote foo)"

[Teemu Likonen <tlikonen@iki.fi>]

On 2009-06-10 04:45 (-0700), florian wrote:

> Dear wizards,

I'm not one but...

> I would not have thought that such an error message was possible,
> since (quote foo) should be the same as 'foo, hence, a symbol. I have
> managed to produce it like this:

Unevaluated "(quote foo)" is not a symbol, it's a list. "foo" is a
symbol.

> (let ((foo 1)
> 	(bar 2)
> 	(fubar 3))
>     (dolist (var '('foo 'bar 'fubar))

Here you have double quoting. You quote the list from being evaluated
"'( ... )" and you have also all the three list items quoted.
Effectively all the list items are lists themselves, that is: "(quote
foo) (quote bar) (quote fubar)".

>       (message "%S's value is %d"
> 	       var (symbol-value var))))

So the value of "var" is a list like "(quote foo)" which is not
evaluated anymore. To fix the problem either drop the double-quoting
(see above) or evaluate "var" once more, like "(eval var)".

> The following form, which should produce the same error, however, does
> not complain (just as I expected):
>
> (let ((var (quote foo)))
>   (symbol-value var))

> (let ((var 'foo))
>   (symbol-value var))

These work because there is no double quoting. The value of "var" is
just symbol "foo", not a list "(quote foo)" like in your previous
example.

> Also, symbolp and symbol-value do not complain if they get sth like
> "(quote foo)" as argument. Can anybody point me to what on earth is
> going on in the first form? Thanks so much!

Actually only in your first, broken example "symbolp" and "symbol-value"
get a list like "(quote foo)" as argument and thus they complain. In
your latter examples they get symbol "foo" as argument. The reason is
the double quoting.