Re: new buffer - should its mode reflect its name when the name matches auto-mode-alist?

[Kevin Rodgers <kevin.d.rodgers@gmail.com>]

Drew Adams wrote:
>>> Haven't thought about this before, but I wonder if the mode 
>>> should respect auto-mode-alist when you create a new buffer?
>>> There's no doubt a good reason or
>>> two why we don't do this - just wondering.
>>>
>>> E.g. `C-x b foo.el', where there is no existing buffer 
>>> foo.el. The mode is
>>> Fundamental; should it be Emacs-Lisp instead?
>>>
>>> (I realize that a new buffer need not be intended to be 
>>> saved as a file.)
>>>
>> As an end-user and not having contributed anything to Emacs myself, I
>> would expect the buffer to "respect" auto-mode-alist, i.e. set the
>> mode depending on the buffer name. However, I think the default
>> behavior when creating buffers non-interactively should *not* respect
>> auto-mode-alist.
> 
> Yes, I meant interactively only, but should have made that clear. Thx.

I think I posted this in response to a request on gnu.emacs.help, back around 
2008-11-18, but I can't find it via Google:

(defadvice switch-to-buffer (around interactive-normal-mode activate)
   "When called interactively to create a new buffer not visiting a file,
temporarily bind `buffer-file-name' and call `normal-mode'."
   (let ((existing-buffer (get-buffer (ad-get-arg 0))))
     ad-do-it
     (when (and (interactive-p)
	       (null existing-buffer)
	       (null buffer-file-name))
       (let ((buffer-file-name (expand-file-name (buffer-name))))
	(normal-mode)))))

Not appropriate for src/buffer.c of course, but you get the idea.

-- 
Kevin Rodgers
Denver, Colorado, USA