Re: 0.97 on xemacs: Symbol's value as variable is void: user-full-name

[Steinar Bang <sb@metis.no>]

>>>>> Steinar Bang <sb@metis.no>:

>>>>> Steinar Bang <sb@metis.no>:
>>>>> Didier Verna <verna@inf.enst.fr>:
>>> Steinar Bang <sb@metis.no> wrote:
>>>> Platform: pgnus 0.97, XEmacs 20.4 w/MULE, BBDB 2.00.3

>>>> I tried upgrading from 0.84 to 0.97, but all commands for sending
>>>> messages (reply, followup and gnus-bug) resulted in the error
>>>> message: Symbol's value as variable is void: user-full-name in the
>>>> minibuffer accompanied by a beep.

>>> Please set debug-on-error to t (Options menu => General Options) and
>>> show us the lisp backtrace.

>> This is all I got (aren't these backtraces usually a lot longer...?):

> Probably because it was byte-compiled.  Here's a backtrace from an
> uncompiled CVS checkout (0.97.1):

> Signaling: (void-variable user-full-name)
>   (or (cdr name) user-full-name)
> )
>   (list (quote user-full-name) (or (cdr name) user-full-name))
> )
>   (list (list (quote user-full-name) (or ... user-full-name)) (list (quote user-mail-address) (or ... user-mail-address)))
> )

Hm... is this this bit (starting at gnus-msg.el:1196)...?
        ...
	(add-hook 'message-setup-hook
		  `(lambda ()
		     (let ((user-full-name ,(or (cdr name) user-full-name))
			   (user-mail-address
			    ,(or (cdr address) user-mail-address)))
		       (save-excursion
			 (message-remove-header "From")
			 (message-goto-eoh)
			 (insert "From: " (message-make-from) "\n")))))))))
        ...

Aren't the things in let clauses sort of local variables?  Isn't it
allowed to overload a function name with a local variable?