[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completio
From: |
martin rudalics |
Subject: |
bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions |
Date: |
Sat, 01 Oct 2011 11:08:09 +0200 |
User-agent: |
Thunderbird 2.0.0.21 (Windows/20090302) |
> 1. emacs -Q
> 2. Type `M-x glo TAB' and at the completion "M-x global-" type TAB again,
> making a *Completions* buffer pop up.
> 3. Type `v TAB', yielding the completion "M-x global-visual-line-mode".
> => The window that had been displaying the *Completions* buffer now
> displays the *Messages* buffer.
>
> AFAICT this is due to the following cond clause in bury-buffer (which is
> called by minibuffer-hide-completions):
>
> ((not (window-dedicated-p))
> (switch-to-prev-buffer nil 'bury))
... plus the fact that switching buffers should not delete windows ...
> Only after typing RET in the minibuffer to accept the completion is the
> window that had been displaying the *Completions* buffer and now
> displays the *Messages* buffer deleted; in Emacs 23.2 it is deleted as
> soon as the sole completion is displayed in the minibuffer. Commenting
> out the above clause restores the 23.2 behavior. But this presumably
> has bad effects elsewhere.
Presumably `minibuffer-hide-completions' should iconify a standalone
completions frame while deleting a split-off completions window. So
maybe the following DTRT:
(defun minibuffer-hide-completions ()
"Get rid of an out-of-date *Completions* buffer."
;; FIXME: We could/should use minibuffer-scroll-window here, but it
;; can also point to the minibuffer-parent-window, so it's a bit tricky.
(let* ((window (get-buffer-window "*Completions*" 0))
(buffer (window-buffer window)))
(when window
(let ((deletable (window-deletable-p window)))
(cond
((eq deletable 'frame)
;; Iconify frame.
(iconify-frame (window-frame window))
(bury-buffer-internal buffer))
(deletable
;; Delete window
(delete-window window)
(bury-buffer-internal buffer))
(t
;; Switch to another buffer.
(switch-to-prev-buffer window 'bury)))))))
> (Interestingly, the C code of bury-buffer prior to the move to Lisp in
> revision 104559 (and long before that) has the equivalent of the above
> clause.)
ISTR that completion windows were dedicated then and typing "v TAB"
would trigger the following code in `minibuffer-completion-help'
;; If there are no completions, or if the current input is already the
;; only possible completion, then hide (previous&stale) completions.
(let ((window (and (get-buffer "*Completions*")
(get-buffer-window "*Completions*" 0))))
(when (and (window-live-p window) (window-dedicated-p window))
(condition-case ()
(delete-window window)
(error (iconify-frame (window-frame window))))))
which I can't find anymore. But I might be wrong.
martin
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions,
martin rudalics <=
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions, Stephen Berman, 2011/10/01
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions, martin rudalics, 2011/10/01
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions, Stefan Monnier, 2011/10/01
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions, martin rudalics, 2011/10/01
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions, Stefan Monnier, 2011/10/01
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions, martin rudalics, 2011/10/02
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions, Stefan Monnier, 2011/10/02
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions, martin rudalics, 2011/10/04
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions, Stefan Monnier, 2011/10/04
- bug#9639: 24.0.90; Problem with bury-buffer in minibuffer-hide-completions, Drew Adams, 2011/10/02