Re: [h-e-w] key-binding: C-s and ESC

[David Vanderschel]

I had written:
>> Interestingly, I have defined C-f in my own
>> isearch-mode-map; but, in my case, it is to add a
>> single character after point to the search string
>> (analogous to C-w, but less greedy).  Even more useful
>> is my C-b which removes a single character (because
>> C-w is sometimes _too_ greedy).

On Friday, November 08, "Nascif Abousalh-Neto" <address@hidden> wrote:
>Your extensions (C-b, C-f) to the isearch mode map
>seem to be quite useful, would you mind posting them?

Yes, actually; I would mind.  I did it by hacking the
isearch-yank function in an older version of
isearch.el (from the version 19 I was running under
OS/2 five years ago).  I introduced isearch-yank-char
and isearch-unyank-char using the hacked isearch-yank
with two new definitions for the "chunk" argument.
The current version of isearch.el does not even seem
to use the old isearch-yank helper function I hacked.
Even though my isearch-yank-char and
isearch-unyank-char do still work correctly for me, I
do not think it appropriate to post such code.

I am also not very proud of how I did it.  It was
simple to make it push the stack for both yank-char
and unyank-char, but that does not really make much
sense.  OTOH, I did not understand well enough how the
code worked in its larger context to modify it in a
more reasonable fashion (nor was I willing to spend
more time trying to figure it out).

If there is someone more familiar with the guts of
isearch who would like to reimplement this more nearly
correctly, I would be happy to share privately the
crude thing I have now.  It works in 20.7.3, but I
cannot guarantee it for 21.  (When I moved over from
OS/2 to Windows 98 about four years ago, I forgot to
byte compile it - and, believe it or not, the .elc
from OS/2 worked correctly with ntemacs!  I believe
that the changes between 19 and 20 which affected .elc
format had been folded into the last '19' version
Eberhard Mattes made for OS/2.)

I do think this is useful additional function for
emacs.  I use it frequently myself.

Regards,
  David V.