If anyone out there is using the Allegro CL Emacs interface, you'll
know that it really kicks some butt.  However, when you're in the
*common-lisp* buffer, like at the top-level, you'll notice that C-a
brings you to the beginning of the line, and that isn't nice when the
prompt is there.  No one wants to edit the prompt!

So here's the fix.  Just put this in your .emacs or wherever.  You can
even put it inside your fi-keys.el file, if you want.  It'll fix the
problem, or at least it did for me:

(defvar acl-prompt-regexp "\\[[0-9]+\\] \\w+([0-9]+): *")

(defun fi:bol ()
  "Newline, but skip passed the prompt, if there is one"
  (interactive)
  (beginning-of-line)
  (when (looking-at acl-prompt-regexp)
    (re-search-forward acl-prompt-regexp)))

(define-key fi:inferior-common-lisp-mode-map [(control a)] 'fi:bol)

* David Bakhash
| If anyone out there is using the Allegro CL Emacs interface, you'll
| know that it really kicks some butt.  However, when you're in the
| *common-lisp* buffer, like at the top-level, you'll notice that C-a
| brings you to the beginning of the line, and that isn't nice when the
| prompt is there.  No one wants to edit the prompt!

  this is the official Emacs design policy.  C-a should not guess a prompt,
  it should move to the beginning of the line in the buffer.  if you want
  to move to the beginning of the text after a prompt, use C-c C-a.  (if
  you want to use the C-c prefix for a number of control characters, set
  the variable FI:SUBPROCESS-ENABLE-SUPERKEYS to true before starting a any
  subprocesses.  see the documentation for the details.)

thanks!

Jeffrey Mark Siskind <q...@research.nj.nec.com> writes:

I note that in a shell buffer, C-a ignores the prompt.  So this
violates Emacs policy.  (Umm, I only tested this on XEmacs.  I don't
have Emacs anymore.)

* Jeffrey Mark Siskind
| I could never understand the logic behind this.  Both a prompt-sensitive
| and a prompt-insensitive move-to-beginning-of-line are useful behaviours.
| However, in my experience, the prompt-sensitive version is used far more
| often than the prompt-insensitive version.  Isn't it desirable to have
| the more-often-used command bound to a shorter keystroke sequence.  I.e.,
| bind C-c C-a to the prompt-insensitive version and bind C-a to the
| prompt-sensitive version?  So I believe that that ILisp got this right
| and the official Emacs policy got this wrong.  Cam someone please
| enlighten me with the compelling reasons that motivate the official Emacs
| policy?

  well, there was a vote on this some time ago, quite obviously because
  this is a question of taste and an application of the principle of least
  astonishment, which I think is subjectivity at its best, and thus an
  arbitrary decision with both good "arguments" and people failing to see
  the logic of the arguments for any given viewpoint.  nonetheless, people
  voted overwhelmingly in favor of C-a retaining its usual semantics.

  now, I don't know whether this is rationalization after the fact or
  actually a good argument in favor of the status quo, but the same can be
  said to apply to other Emacs editing commands that have dual meaning in
  various special modes, like C-c, C-d, C-w, C-z, etc.