Menu
▾
▴
#2 problems with completer from emacs 21.1
When using ilisp.emacs (from version 5.11) as Emacs startup file (only the path to the ilisp files was changed) the Emacs minibuffer completion stopped working.
When removing (require 'completer) from .emacs, this effect is delayed until one does 'run-ilisp".
XEmacs 21.1 (patch 14) works fine with the same .emacs file.
Logged In: NO
I'm able to reproduce this problem with
GNU Emacs 21.1.1 (i386-msvc-nt5.0.2195) of 2001-10-22
and ilisp 5.11.
Logged In: YES
user_id=38855
This is fixed in current CVS ILISP.
Logged In: NO
with ilisp 5.11, you can work around this problem by using
this directive in your .emacs file:
(setq ilisp-*prefix-match* t)
Tord Kallqvist Romstad pointed this out in a posting to
comp.lang.lisp.
Logged In: YES
user_id=275595
I don't know if this helps, but I found trouble with the
completion of symbols, and tracked it down to a bug in
minibuffer.
I'm not sure whether this is the same problem. Here's some
information I sent to the xemacs maintainers:
I BELIEVE that ilisp-cmp has a bug in ilisp-read-symbol,
particularly
in the way it uses the minibuffer. I was finding that this
crashed in
my copy of ilisp (ilisp-cmp.el,v 1.4) on Xemacs 21.1.14, linux.
When using meta-dot to find a function definition, I was getting
completion crashes with the error that LAST-EXACT-COMPLETION was
unbound.
That variable is defined in minibuf.el (no version number
available in
my copy :-<), but not bound.
When one calls SOME minibuffer completion function, it seems
to behave in
a well-balanced way: the function first binds
last-exact-completion
to nil, then starts functioning.
However, if one calls minibuffer-complete, as
completer-complete in
ilisp-read-symbol does, last-exact-completion is not bound.
The head of minibuffer-do-completion has the following in it:
(defun minibuffer-do-completion (buffer-string)
(let* ((completion (try-completion buffer-string
minibuffer-completion-table
minibuffer-completion-predicate))
(status (minibuffer-do-completion-1 buffer-string
completion))
(last last-exact-completion))
...note that this assumes last-exact-completion is bound on
entry to
the function.
Since minibuffer-complete is interactive, and thus can be called
directly, I BELIEVE that the appropriate patch is to modify
minibuffer-complete as follows:
OLD version:
(defun minibuffer-complete ()
"Complete the minibuffer contents as far as possible.
Return nil if there is no valid completion, else t.
If no characters can be completed, display a list of
possible completions.
If you repeat this command after it displayed such a list,
scroll the window of possible completions."
(interactive)
;; If the previous command was not this, then mark the
completion
;; buffer obsolete.
(or (eq last-command this-command)
(setq minibuffer-scroll-window nil))
(let ((window minibuffer-scroll-window))
(if (and window (windowp window) (window-buffer window)
(buffer-name (window-buffer window)))
;; If there's a fresh completion window with a live buffer
;; and this command is repeated, scroll that window.
(let ((obuf (current-buffer)))
(unwind-protect
(progn
(set-buffer (window-buffer window))
(if (pos-visible-in-window-p (point-max) window)
;; If end is in view, scroll up to the beginning.
(set-window-start window (point-min))
;; Else scroll down one frame.
(scroll-other-window)))
(set-buffer obuf))
nil)
(let ((status (minibuffer-do-completion (buffer-string))))
(if (eq status 'none)
nil
(progn
(cond ((eq status 'unique)
(temp-minibuffer-message
" [Sole completion]"))
((eq status 'exact)
(temp-minibuffer-message
" [Complete, but not unique]")))
t))))))
NEW version:
(defun minibuffer-complete ()
"Complete the minibuffer contents as far as possible.
Return nil if there is no valid completion, else t.
If no characters can be completed, display a list of
possible completions.
If you repeat this command after it displayed such a list,
scroll the window of possible completions."
(interactive)
;; If the previous command was not this, then mark the
completion
;; buffer obsolete.
(cond ((not (eq last-command this-command))
(setq minibuffer-scroll-window nil)
(setq last-exact-completion nil)))
(let ((window minibuffer-scroll-window))
(if (and window (windowp window) (window-buffer window)
(buffer-name (window-buffer window)))
;; If there's a fresh completion window with a live buffer
;; and this command is repeated, scroll that window.
(let ((obuf (current-buffer)))
(unwind-protect
(progn
(set-buffer (window-buffer window))
(if (pos-visible-in-window-p (point-max) window)
;; If end is in view, scroll up to the beginning.
(set-window-start window (point-min))
;; Else scroll down one frame.
(scroll-other-window)))
(set-buffer obuf))
nil)
(let ((status (minibuffer-do-completion (buffer-string))))
(if (eq status 'none)
nil
(progn
(cond ((eq status 'unique)
(temp-minibuffer-message
" [Sole completion]"))
((eq status 'exact)
(temp-minibuffer-message
" [Complete, but not unique]")))
t))))))
Logged In: YES
user_id=38855
Can you please test with a recent ILISP version?
ILISP 5.11 completer.el is known to be broken with Emacs 21.
We fixed the problem in CVS in or before February 2002.
Since then, ILISP 5.12 and a snapshot have been released.
Logged In: YES
user_id=275595
I'm not sure that mine is really the same bug as 487717. I
just checked, and the version of ilisp-cmp.el that is in the
current ilisp and the version in mine (the one that's
distributed from xemacs.org) are identical, except for the
version number.
And, anyway, the bug I found is in the minibuffer code, not
in ilisp, so nothing in the ilisp distribution could fix it.
Logged In: YES
user_id=275595
I'm not sure that mine is really the same bug as 487717. I
just checked, and the version of ilisp-cmp.el that is in the
current ilisp and the version in mine (the one that's
distributed from xemacs.org) are identical, except for the
version number.
And, anyway, the bug I found is in the minibuffer code, not
in ilisp, so nothing in the ilisp distribution could fix it.