Re: PATCH: isearch-yank-until-char

[Karl Fogel <kfogel@red-bean.com>]

On 14 Aug 2019, Juri Linkov wrote:
>Globally 'C-M-c' is bound to 'exit-recursive-edit', so the question is
>does anyone want 'C-M-c' to quit isearch and execute 'exit-recursive-edit'?
>I never used 'C-M-c' to quit isearch.

Neither have I.

>One problem is that typing 'C-s C-M-c <SPC>' at the end of the buffer
>raises the error:
>
>  Debugger entered--Lisp error: (search-failed " ")
>    search-forward(" ")
>    isearch-yank-until-char(32)
>    funcall-interactively(isearch-yank-until-char 32)
>    call-interactively(isearch-yank-until-char nil nil)
>    command-execute(isearch-yank-until-char)
>
>and leaves isearch in a broken state.  Should your new command
>catch this error?

It could.  I'm not sure exacly what to do with the error, though.  The new version below displays a brief message and then returns to the isearch prompt:

(defun isearch-yank-until-char (char)
  "Pull everything until next instance of CHAR from buffer into search string.
Interactively, prompt for CHAR."
  (interactive "cYank until character: ")
  (isearch-yank-internal
   (lambda () (let ((inhibit-field-text-motion t))
                (condition-case nil
                    (progn
                      (search-forward (char-to-string char))
                      (forward-char -1))
                  (search-failed
                   (message "`%c' not found" char)
                   (sit-for 2)))
                (point)))))

This feels klugey, but I'm not sure what a better way is.  Do you have any ideas?

I also tried raising an error -- i.e., substitute `message' above with `error' and get rid of the `(sit-for 2)' -- but the problem with that is that it *never* returns to the isearch prompt, even though in either case isearch is still active and one could continue adding to the search string.

Best regards,
-Karl