bug#10654: 24.0.93; isearch-yank-line moved from C-y

bug#10654: 24.0.93; isearch-yank-line moved from C-y

[Aaron S. Hawley]

I appreciate the new behavior of M-y in Isearch Mode.  I find it more
satisfying, and less frustrating, to be able to access the full kill
ring.

I was surprised that the previously unsatisfying command M-y overwrote
the behavior of C-y in Isearch Mode.  C-y had held the very useful
command of yanking the rest of the line into Isearch.  I eat and
breathe Diff output and line-oriented source code, so I would use the
C-y command often to search similar text or compare text in the buffer
or across buffers.  I use M-x diff or M-x compare-windows but Isearch
Mode can get the job done quicker in some cases.

I don't have any data to support that C-y is popular.  I won't suggest
that the bindings for Isearch Mode are anywhere near sane.  I recall I
was flummoxed by them at first.  The bindings are diabolical for
beginners in most ways, but how did I learn them if they weren't
coherent?  However, the new change is disruptive for existing users of
Emacs and after trying it for two weeks, I can confirm that putting
`isearch-yank-line' to the `M-s C-e' binding is unsatisfying (M-s C-k
makes more sense).  Though, without any data maybe I'm the only person
who uses it and needs a short key binding?

The proposed bindings for 24.1 are:

(define-key isearch-mode-map "\C-y" 'isearch-yank-kill)
(define-key isearch-mode-map "\M-y" 'isearch-yank-pop)
(define-key isearch-mode-map "\M-s\C-e" 'isearch-yank-line)

I'd prefer to revert partially to:

(define-key isearch-mode-map "\C-y" 'isearch-yank-line)
(define-key isearch-mode-map "\M-y" 'isearch-yank-pop)

Or keep the proposed new bindings, but at least give a better binding
than M-s C-e  for the old C-y.  How about the shorter (and obvious)
key binding of C-k?

(define-key isearch-mode-map "\C-y" 'isearch-yank-kill)
(define-key isearch-mode-map "\M-y" 'isearch-yank-pop)
(define-key isearch-mode-map "\C-k" 'isearch-yank-line)

Thanks for Emacs,
aaron

-- 
In general, we reserve the right to have a poor
memory--the computer, however, is supposed to
remember!  Poor computer.  -- Guy Lewis Steele Jr.

bug#10654: 24.0.93; isearch-yank-line moved from C-y

[Glenn Morris]

"Aaron S. Hawley" wrote:

> I don't have any data to support that C-y is popular.

The weight of opinion seems to be against you.

http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7700#11
  I've always found the C-y binding in isearch "useless" and
  counter-intuitive. I'd much rather make C-y and M-y behave like they
  do in the minibuffer.

http://debbugs.gnu.org/cgi/bugreport.cgi?bug=8183#5
  While I'm sure there's someone in the world who has a use for
  searching for the entire next line, there are many, many more people
  who find it useful to search for items from the kill ring.

http://lists.gnu.org/archive/html/emacs-devel/2008-11/msg00687.html
  I was reluctant about it in the first place.

> I can confirm that putting `isearch-yank-line' to the `M-s C-e'
> binding is unsatisfying (M-s C-k makes more sense). 

Here is the motivation:
http://lists.gnu.org/archive/html/emacs-devel/2008-11/msg00606.html
   bind `isearch-yank-line' to `M-s C-e' because `move-end-of-line' is
   on `C-e'.

> Though, without any data maybe I'm the only person who uses it and
> needs a short key binding?
[...]
> How about the shorter (and obvious) key binding of C-k?
[...]
> (define-key isearch-mode-map "\C-k" 'isearch-yank-line)

Personally I have no opinion.

bug#10654: 24.0.93; isearch-yank-line moved from C-y

[Aaron S. Hawley]

On 1/30/12, Glenn Morris <rgm@gnu.org> wrote:
> "Aaron S. Hawley" wrote:
>
>> I don't have any data to support that C-y is popular.
>
> The weight of opinion seems to be against you.

This is what I expected.

Thanks for tracking down the threads, Glenn.

It would seem the binding for hitting C-y in Isearch to yank the rest
of the line existed ever since Daniel LaLiberte wrote Isearch 20 years
ago.  I guess its time has come.  This means hitting C-k C-/ C-s C-y
rather than C-s C-y.  I can relinquish that C-y in Isearch wasn't
consistent with global C-y in Emacs, but if word and character yanking
are on C-w and C-M-y, respectively, it seems like line-yanking should
be on a better binding than M-s C-e.

Thanks again,
aaron

> http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7700#11
>   I've always found the C-y binding in isearch "useless" and
>   counter-intuitive. I'd much rather make C-y and M-y behave like they
>   do in the minibuffer.
>
> http://debbugs.gnu.org/cgi/bugreport.cgi?bug=8183#5
>   While I'm sure there's someone in the world who has a use for
>   searching for the entire next line, there are many, many more people
>   who find it useful to search for items from the kill ring.
>
> http://lists.gnu.org/archive/html/emacs-devel/2008-11/msg00687.html
>   I was reluctant about it in the first place.
>
>> I can confirm that putting `isearch-yank-line' to the `M-s C-e'
>> binding is unsatisfying (M-s C-k makes more sense).
>
> Here is the motivation:
> http://lists.gnu.org/archive/html/emacs-devel/2008-11/msg00606.html
>    bind `isearch-yank-line' to `M-s C-e' because `move-end-of-line' is
>    on `C-e'.

-- 
In general, we reserve the right to have a poor
memory--the computer, however, is supposed to
remember!  Poor computer.  -- Guy Lewis Steele Jr.

bug#10654: 24.0.93; isearch-yank-line moved from C-y

[Juri Linkov]

> (define-key isearch-mode-map "\C-k" 'isearch-yank-line)

To tell the truth, this is exactly what I currently use in .emacs.

`C-k' is a very convenient binding for `isearch-yank-line',
but I hesitated to propose it for adding to isearch.el
because I expect the opposition to that.

bug#10654: 24.0.93; isearch-yank-line moved from C-y

[Kevin Rodgers]

On 1/30/12 1:19 PM, Aaron S. Hawley wrote:

> It would seem the binding for hitting C-y in Isearch to yank the rest
> of the line existed ever since Daniel LaLiberte wrote Isearch 20 years
> ago.  I guess its time has come.  This means hitting C-k C-/ C-s C-y
> rather than C-s C-y.  I can relinquish that C-y in Isearch wasn't
> consistent with global C-y in Emacs, but if word and character yanking
> are on C-w and C-M-y, respectively, it seems like line-yanking should
> be on a better binding than M-s C-e.

Or word and character yanking could be bound consistently: M-s M-f and M-s C-f

That approach would be a little less onerous if additional words or characters
could be added to the search string by additional M-f and C-f without prefixing
each with M-s.

-- 
Kevin Rodgers
Denver, Colorado, USA

bug#10654: 24.0.93; isearch-yank-line moved from C-y

[Juri Linkov]

> That approach would be a little less onerous if additional words or characters
> could be added to the search string by additional M-f and C-f without prefixing
> each with M-s.

If `C-k' for `isearch-yank-line' is not under consideration,
then I recommend to allow `M-s C-e C-e C-e ...' in 24.1
because `M-s C-e M-s C-e M-s C-e ...' is unusable.

This can be achieved with the patch below.

Other repeating key sequences `M-s M-f M-f M-f ...' and `M-s C-f C-f C-f ...'
could be implemented as well, but there is no urgent need for them now.

=== modified file 'lisp/isearch.el'
--- lisp/isearch.el	2012-02-01 00:21:17 +0000
+++ lisp/isearch.el	2012-02-04 00:17:07 +0000
@@ -533,6 +533,9 @@ (defvar minibuffer-local-isearch-map
     map)
   "Keymap for editing Isearch strings in the minibuffer.")
 
+(defvar isearch-repeat-key nil)
+(defvar isearch-repeat-command nil)
+
 ;; Internal variables declared globally for byte-compiler.
 ;; These are all set with setq while isearching
 ;; and bound locally while editing the search string.
@@ -1662,6 +1665,8 @@
 (defun isearch-yank-line ()
   "Pull rest of line from buffer into search string."
   (interactive)
+  (setq isearch-repeat-key last-input-event
+	isearch-repeat-command 'isearch-yank-line)
   (isearch-yank-internal
    (lambda () (let ((inhibit-field-text-motion t))
 		(line-end-position (if (eolp) 2 1))))))
@@ -1962,8 +1967,13 @@ (defun isearch-other-meta-char (&optiona
                 (this-command-keys)))
 	 (main-event (aref key 0))
 	 (keylist (listify-key-sequence key))
+	 (repeat-key isearch-repeat-key)
          scroll-command isearch-point)
-    (cond ((and (= (length key) 1)
+    (setq isearch-repeat-key nil)
+    (cond ((and (eq main-event repeat-key) isearch-repeat-command)
+	   (setq isearch-repeat-key main-event)
+	   (command-execute isearch-repeat-command))
+	  ((and (= (length key) 1)
 		(let ((lookup (lookup-key local-function-key-map key)))
 		  (not (or (null lookup) (integerp lookup)
 			   (keymapp lookup)))))

bug#10654: 24.0.93; isearch-yank-line moved from C-y

[Lars Ingebrigtsen]

Glenn Morris <rgm@gnu.org> writes:

>> I don't have any data to support that C-y is popular.
>
> The weight of opinion seems to be against you.

(I'm going through old bug reports that unfortunately weren't resolved
at the time.)

I think the conclusion here is that we didn't want to change these key
bindings, so I'm closing this bug report.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no