"(setq name)": Is this a bug?

"(setq name)": Is this a bug?

[Alan Mackenzie]

Hi, Emacs,

In .../lisp/files.el, function set-auto-mode, L2553, appears the strange
invocation:

    (setq name)

.  One of my analysis tools tripped up on it.  According to setq's doc
string, this is invalid.  In each [SYM VAL] in (setq [SYM VAL] ...), the
VAL is not optional.  Surely?

I think this is a bug.

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: "(setq name)": Is this a bug?

[Ian Eure]

On Jul 14, 2009, at 9:01 AM, Alan Mackenzie wrote:

> Hi, Emacs,
>
> In .../lisp/files.el, function set-auto-mode, L2553, appears the  
> strange
> invocation:
>
>    (setq name)
>
> .  One of my analysis tools tripped up on it.  According to setq's doc
> string, this is invalid.  In each [SYM VAL] in (setq [SYM VAL] ...),  
> the
> VAL is not optional.  Surely?
>
I suspect the documentation just isn't clear. From a brief reading of  
the source, setq supports 0/1/odd arguments, with `nil' being used in  
the case of non-mod-two argument arity.

In a nutshell, it loops over the arguments two at a time until they're  
empty, setting the car of the args_left to the evaluated cadr of  
args_left. If you're at the end of a list, the cdr is `nil', and the  
car of `nil' is also `nil'.

So if you give it an odd number of arguments, the last value will be  
nil.

  - Ian

Re: "(setq name)": Is this a bug?

[Stefan Monnier]

> Hi, Emacs,
> In .../lisp/files.el, function set-auto-mode, L2553, appears the strange
> invocation:

>     (setq name)

> .  One of my analysis tools tripped up on it.  According to setq's doc
> string, this is invalid.  In each [SYM VAL] in (setq [SYM VAL] ...), the
> VAL is not optional.  Surely?

> I think this is a bug.

It's accepted bny the interpreter and the compiler, AFAIK, but I do
consider it as a bug, so please fix it to (setq name nil).


        Stefan