kill-region

kill-region

[Luc Teirlinck]

kill-region sets this-command to 'kill-region even if the last command
was not a kill command *and* kill-region did not make a new entry on
the kill-ring (which happens if kill-region is called on an empty
region).

This can lead to strange consequences.  Start out with some non-empty
kill ring.  Go to the end of some buffer containing at least two lines.
Do M-2 k.  Nothing got killed, since we are at the end of the buffer.
Now do M--2 k. (What we probably wanted to do, we just forgot the
"-".)  Now C-y.  The last two lines of the buffer got prepended to the
previous kill ring entry.  Does this make sense?  We could have
avoided the mess by doing a C-g after the erroneous M-2 k, but it
still seems very counter-intuitive.

Might it be better for kill-region to either not set this-command to
'kill-region if the last command was not already a kill command and
kill-region did not make a new entry on the kill ring, or,
alternatively, make kill-region add an empty string to the end of the
kill ring in the described situation, to which subsequent kill
commands could append or prepend?

Sincerely,

Luc.

kill-region

[Luc Teirlinck]

Fom my previous message on the topic:

   Do M-2 k.  Nothing got killed, since we are at the end of the buffer.
   Now do M--2 k.

Meant was M-2 C-k and M--2 C-k.

   We could have avoided the mess by doing a C-g after the erroneous
   M-2 k,

M-2 C-k again.

   or, alternatively, make kill-region add an empty string to the end
   of the kill ring

Meant was, of course, the beginning of the kill ring.

Sorry for all these typos.

Sincerely,

Luc.

Re: kill-region

[Kai Großjohann]

Luc Teirlinck <teirllm@dms.auburn.edu> writes:

> Might it be better for kill-region to either not set this-command to
> 'kill-region if the last command was not already a kill command and
> kill-region did not make a new entry on the kill ring, or,
> alternatively, make kill-region add an empty string to the end of the
> kill ring in the described situation, to which subsequent kill
> commands could append or prepend?

I guess it's better to avoid empty strings on the kill ring, if
possible.

I can't say what might be the consequences of frobbing this-command.
-- 
This line is not blank.

Re: kill-region

[Richard Stallman]

    Might it be better for kill-region to either not set this-command to
    'kill-region if the last command was not already a kill command and
    kill-region did not make a new entry on the kill ring, or,
    alternatively, make kill-region add an empty string to the end of the
    kill ring in the described situation, to which subsequent kill
    commands could append or prepend?

I agree that we should do one or the other.  I think the former is
better, since I don't see much use in putting the empty region in the
kill ring.  As long as we can provide fairly consistent behavior
while not putting the empty region in the kill ring, let's do so.

Re: kill-region

[Kai Großjohann]

Richard Stallman <rms@gnu.org> writes:

>     Might it be better for kill-region to either not set this-command to
>     'kill-region if the last command was not already a kill command and
>     kill-region did not make a new entry on the kill ring, or,
>     alternatively, make kill-region add an empty string to the end of the
>     kill ring in the described situation, to which subsequent kill
>     commands could append or prepend?
>
> I agree that we should do one or the other.  I think the former is
> better, since I don't see much use in putting the empty region in the
> kill ring.  As long as we can provide fairly consistent behavior
> while not putting the empty region in the kill ring, let's do so.

How about this?

--- simple.el.~1.604.~	Thu May 22 09:59:08 2003
+++ simple.el	Fri May 23 21:40:41 2003
@@ -1913,7 +1913,8 @@
 	  (if (eq last-command 'kill-region)
 	      (kill-append string (< end beg) yank-handler)
 	    (kill-new string nil yank-handler)))
-	(setq this-command 'kill-region))
+	(when (or string (eq last-command 'kill-region))
+	  (setq this-command 'kill-region)))
     ((buffer-read-only text-read-only)
      ;; The code above failed because the buffer, or some of the characters
      ;; in the region, are read-only.

-- 
This line is not blank.