fset and error reporting

fset and error reporting

[Dmitry Antipov]

My .emacs has:

(fset 'yes-or-no-p 'y-or-n-p)

With -Q, I have:

(yes-or-no-p) ==> Wrong number of arguments: yes-or-no-p, 0

After loading .emacs, I have mysterious:

(yes-or-no-p) ==> Wrong number of arguments: (1 . 1), 0

Can someone explain why it is so?

Dmitry

Re: fset and error reporting

[Jordon Biondo]

It does not have to do with fset. Rather yes-or-no-p is a `subrp` and is being evaluated by  eval.c(eval_sub) and y-or-no-p is a `byte-code-function-p` evaluated by bytecode.c(exec_byte_code). These two functions both signal about incorrect number of args if needed, but pass different data in. bytecode.c seems to be the oddball as all such signals pass function name and number of args given but bytecode.c passes a cons with min/max args.

If you’re actually wondering why it was chosen that sometimes the cons cell is passed in and sometimes not, well I have no idea.

Hope it helps!

Jordon

On Sep 2, 2014, at 1:33 PM, Dmitry Antipov <dmantipov@yandex.ru> wrote:

> My .emacs has:
> 
> (fset 'yes-or-no-p 'y-or-n-p)
> 
> With -Q, I have:
> 
> (yes-or-no-p) ==> Wrong number of arguments: yes-or-no-p, 0
> 
> After loading .emacs, I have mysterious:
> 
> (yes-or-no-p) ==> Wrong number of arguments: (1 . 1), 0
> 
> Can someone explain why it is so?
> 
> Dmitry
>