bug#21208: 25.0.50; isearch gets confused with 'too many words'

bug#21208: 25.0.50; isearch gets confused with 'too many words'

[Sebastien Vauban]

Dima Kogan <dima-8Bn8oVo35dduSJGsrdOfxA@public.gmane.org> writes:
> Hi. I'm running a very recent build of emacs from git:
>
>   https://github.com/emacs-mirror/emacs/commit/0aec2aaccd8b745fa7214f3edd453c04a04bfba4
>
> I just found another isearch regression. Recipe:
>
> 1. emacs -Q
>
> 2. Create or open file such that the buffer contains
> -------------------------
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
>
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
> -------------------------
> i.e two identical long lines containing lots of words '1234'.
>
> 3. Point to start of buffer
>
> 4. C-s
>
> 5. C-w multiple times. As you press C-w, you grow the string being
> sought; you can see the match highlighted on the line being selected and
> on the next line
>
> 4. Eventually (column 239 or so) the minibuffer says 'too many words',
> and the search no longer matches anything on either line. The sought
> string looks like it's appending words without spaces ("123412341234"
> instead of "1234 1234 1234"). Manually fixing that with M-e doesn't
> appear to help.

Confirmed -- was about to report it as well!

Best regards,
  Seb

bug#21208: 25.0.50; isearch gets confused with 'too many words'

[Dima Kogan]

Hi. I'm running a very recent build of emacs from git:

  https://github.com/emacs-mirror/emacs/commit/0aec2aaccd8b745fa7214f3edd453c04a04bfba4

I just found another isearch regression. Recipe:

1. emacs -Q

2. Create or open file such that the buffer contains
-------------------------
1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234

1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234 1234
-------------------------
i.e two identical long lines containing lots of words '1234'.

3. Point to start of buffer

4. C-s

5. C-w multiple times. As you press C-w, you grow the string being
sought; you can see the match highlighted on the line being selected and
on the next line

4. Eventually (column 239 or so) the minibuffer says 'too many words',
and the search no longer matches anything on either line. The sought
string looks like it's appending words without spaces ("123412341234"
instead of "1234 1234 1234"). Manually fixing that with M-e doesn't
appear to help.

bug#21208: 25.0.50; isearch gets confused with 'too many words'

[Juri Linkov]

> 6. Eventually (column 239 or so) the minibuffer says 'too many words',
> and the search no longer matches anything on either line. The sought
> string looks like it's appending words without spaces ("123412341234"
> instead of "1234 1234 1234"). Manually fixing that with M-e doesn't
> appear to help.

A possible solution in case of the error “Too many words” is to fall back
to the ordinary search, and continue the search with success:

diff --git a/lisp/isearch.el b/lisp/isearch.el
index 8d4bf24..a888f79 100644
--- a/lisp/isearch.el
+++ b/lisp/isearch.el
@@ -2720,6 +2734,10 @@ (defun isearch-search ()
       ((and (not isearch-regexp)
 	    (string-match "\\`Regular expression too big" isearch-error))
        (cond
+        ((eq isearch-word #'character-fold-to-regexp)
+         ;; Fallback to an ordinary search
+         (setq isearch-word nil)
+         (isearch-search))
 	(isearch-word
 	 (setq isearch-error "Too many words"))
 	((and isearch-lax-whitespace search-whitespace-regexp)

bug#21208: 25.0.50; isearch gets confused with 'too many words'

[Noam Postavsky]

tag 21208 + unreproducible
quit

I can't reproduce this in 25.0.94, is it fixed?