bug#14268: 24.3.50; isearch-repeat-forward without isearch-forward first signals error

bug#14268: 24.3.50; isearch-repeat-forward without isearch-forward first signals error

[Darren Hoo]

Start Emacs cleanly and doing an isearch-repeat-forward, it is
expected to show the message [No previous search string] in the
echo area, but it raises the following instead:

Debugger entered--Lisp error: (wrong-type-argument arrayp nil)
  isearch-fail-pos(t)
  isearch-message()
  isearch-update()
  isearch-repeat(forward)
  isearch-repeat-forward()
  call-interactively(isearch-repeat-forward nil nil)
  command-execute(isearch-repeat-forward)


In GNU Emacs 24.3.50.1 (x86_64-apple-darwin12.3.0, NS apple-appkit-1187.37)
 of 2013-04-25 on Darren-rMBP.local
Bzr revision: 112377 yamaoka@jpl.org-20130424220721-t957vh44sdvmxj28
Windowing system distributor `Apple', version 10.3.1187
Configured using:
 `configure --with-ns'

diff --git a/lisp/isearch.el b/lisp/isearch.el
index b36b250..b38bfc8 100644
--- a/lisp/isearch.el
+++ b/lisp/isearch.el
@@ -1106,8 +1106,9 @@ If MSG is non-nil, use variable
`isearch-message', otherwise `isearch-string'."
        (curr-msg (if msg isearch-message isearch-string))
        succ-msg)
     (when (or (not isearch-success) isearch-error)
-      (while (or (not (isearch--state-success (car cmds)))
-                 (isearch--state-error (car cmds)))
+      (while (and cmds
+                 (or (not (isearch--state-success (car cmds)))
+                     (isearch--state-error (car cmds))))
         (pop cmds))
       (setq succ-msg (and cmds (if msg (isearch--state-message (car cmds))
                                 (isearch--state-string (car cmds)))))

bug#14268: 24.3.50; isearch-repeat-forward without isearch-forward first signals error

[Darren Hoo]

Sorry, I tried this over and over but can not reproduce this 100% of the
time.

the isearch key binding `C-s' alone is all I need, what's the usage of
`s-d' and `s-g'?

why isearch-repeat-forward and isearch-repeat-backward is bound to the
keys s-d and s-g? I can not find where they're bound(start with Emacs
-Q),

bug#14268: 24.3.50; isearch-repeat-forward without isearch-forward first signals error

[Juri Linkov]

> Sorry, I tried this over and over but can not reproduce this 100% of the
> time.
>
> the isearch key binding `C-s' alone is all I need, what's the usage of
> `s-d' and `s-g'?

Perhaps `s-d' is defined somewhere in your config with something like:

  (define-key global-map [(super ?d)] 'isearch-repeat-forward)

But I can't reproduce the problem by typing `s-d s-d s-d ...'
It displays "[No previous search string]" as expected.

bug#14268: 24.3.50; isearch-repeat-forward without isearch-forward first signals error

[Darren Hoo]

Juri Linkov <juri@jurta.org> writes:


> Perhaps `s-d' is defined somewhere in your config with something like:
>
>   (define-key global-map [(super ?d)] 'isearch-repeat-forward)
>
> But I can't reproduce the problem by typing `s-d s-d s-d ...'
> It displays "[No previous search string]" as expected.

Have you bound these keys in your config? I don't, even if I did I've
already started test with Emacs -Q and it is bound:

   s-d runs the command isearch-repeat-backward, which is an interactive
   compiled Lisp function in `isearch.el'.

I think it's the default key binding, but after grepped a lot from
nowhere can I find  the key `s-d' bound to isearch-repeat-backward.

bug#14268: 24.3.50; isearch-repeat-forward without isearch-forward first signals error

[Juri Linkov]

> I think it's the default key binding, but after grepped a lot from
> nowhere can I find  the key `s-d' bound to isearch-repeat-backward.

Grepping in the source directory `lisp' finds these ns-specific lines:

./lisp/term/ns-win.el:119:(define-key global-map [?\s-d] 'isearch-repeat-backward)
./lisp/term/ns-win.el:122:(define-key global-map [?\s-g] 'isearch-repeat-forward)

One way to reproduce your bug report is before typing `s-g'
set `isearch-error' to a non-nil value, e.g.

  (setq isearch-error "error")

I don't know what code in your config does this, so I installed
your patch as a precaution against such situation.  Thank you.