On Wed, 13 Jul 2011 11:25:10 -0400, rvclayton@verizon.net (R. Clayton) wrote:
> I am running GNU Emacs 23.3.1 (i486-pc-linux-gnu, GTK+ Version 2.24.3) of
> 2011-04-10 on raven, modified by Debian on a debian testing system updated
> weekly.
> 
> I have noticed that the initial scratch buffer, the one you get when you just
> run emacs with no command-line arguments, is in fundamental mode, even though
> .emacs contains, in custom-set-variables, '(initial-major-mode (quote
> text-mode)) and help-variable in *scratch* reports initial-major-mode's value
> is text-mode.

For me, using a ~/.emacs containing only this, works as expected:

(custom-set-variables
  ;; custom-set-variables was added by Custom.
  ;; If you edit it by hand, you could mess it up, so be careful.
  ;; Your init file should contain only one such instance.
  ;; If there is more than one, they won't work right.
 '(initial-major-mode (quote text-mode)))

So, your problem might be caused by something else in your configuration.

> I have also noticed that doing (setq major-mode 'text-mode) in .emacs has no
> effect: new buffers are set to fundamental mode rather than text mode.  The
> help-variable documentation for major-mode indicates that setting help-mode
> makes it buffer local, which seems like strange behavior for a variable that's
> supposed to provide a global value. [..]

major-mode is not supposed to provide a global value. The idea is that
every buffer has its own major-mode, otherwise you couldn't have buffers
open for both C and lisp files and have them highlighted differently for
example. So it makes perfect sense for major-mode to be buffer-local.

If you want to set a default for a buffer-local variable, you can use
setq-default, like so:

(setq-default major-mode 'text-mode)

But because some functions that create a new buffers directly set the
major-mode of the buffer (e.g. mail clients creating a buffer to display
a message), it is doubtful what that line would actually accomplish.

> 
> Although I expect I know the answer to this one, I'll ask it anyway: why is it
> that a "top-level" setq on major-mode in .emacs doesn't work?

Since major-mode is buffer-local, that only changes the major-mode for
the currently selected buffer when that code is executed (depending on
your configuration, that might be the Emacs welcome buffer).

> Also, top-level setqs on initial-major-mode and major-mode used to work as
> expected up until a few weeks ago.  What has changed since then?  Searching
> around in the /usr/share/doc/emacs23 change files doesn't produce anything I
> can recognize as an explanation.

What exactly do you mean by work as expected? If you mean that (setq
major-mode 'text-mode) used to set the major-mode for the
scratch-buffer, then this only could have worked if *scratch* was the
current buffer when .emacs was run. I think this is an implementation
detail that you should not rely on anyway.

About initial-major-mode not working: As I said, I can't reproduce
that. (I'm also using Emacs 23.3.1, but from debian unstable).

Cheers,
Daniel