bug#12677: 24.2.50; `nreverse' complains that ("foo") is not `listp'

bug#12677: 24.2.50; `nreverse' complains that ("foo") is not `listp'

[Drew Adams]

I do not understand this, and I haven't found a way to repro it from
emacs -Q.  (I can give a much longer recipe, but I doubt you are
interested.)
 
Here is the backtrace.  The first line makes it clear that the list
("c:/drews-lisp-20/CONTRIB/") is not considered `listp'.  But if you
just start with (setq dir-names '("c:/drews-lisp-20/CONTRIB/")) there is
no problem.  Yet `e dir-names' shows that that is the value.  And `e
(listp dir-names)' returns t.  Bizarre.  Dunno whether this report will
help in any way, but this debugger behavior sure seems odd.
 
Debugger entered--Lisp error: (wrong-type-argument listp ("c:/drews-lisp-20=
/CONTRIB/"))
  nreverse(("c:/drews-lisp-20/CONTRIB/"))
  (setq dir-names (nreverse (delete "" dir-names)))
  (prog1 (setq dir-names (nreverse (delete "" dir-names))) (icicle-unbind-f=
ile-candidate-keys) (if (interactive-p) (progn (message "Directories: %S" d=
ir-names))))
  (let* ((icicle-orig-buff (current-buffer)) (icicle-orig-window (selected-=
window)) (prompt (or icicle-prompt "Choose directory (`RET' when done): "))=
 (dir-names nil) (icicle-proxy-candidates (let ((ipc nil)) (if icicle-add-p=
roxy-candidates-flag (progn (setq ipc (icicle-remove-if-not ... ...) ipc (a=
pply ... ipc) ipc (icicle-remove-if-not ... ipc)))) ipc)) (user-file-pred i=
cicle-file-predicate) (icicle-file-predicate (if user-file-pred (function (=
lambda (f) (and (file-directory-p f) (funcall user-file-pred f)))) (functio=
n file-directory-p))) (icicle-comp-base-is-default-dir-p t) (completion-ign=
ore-case (or (and (boundp (quote read-file-name-completion-ignore-case)) re=
ad-file-name-completion-ignore-case) completion-ignore-case)) (icicle-show-=
Completions-initially-flag (or icicle-show-Completions-initially-flag icicl=
e-files-ido-like-flag)) (icicle-top-level-when-sole-completion-flag (or ici=
cle-top-level-when-sole-completion-flag icicle-files-ido-like-flag)) (icicl=
e-default-value (if (and icicle-files-ido-like-flag icicle-default-value) i=
cicle-files-ido-like-flag (and (memq icicle-default-value (quote (t nil))) =
icicle-default-value))) (icicle-must-match-regexp icicle-file-match-regexp)=
 (icicle-must-not-match-regexp icicle-file-no-match-regexp) (icicle-must-pa=
ss-after-match-predicate icicle-file-predicate) (icicle-require-match-flag =
icicle-file-require-match-flag) (icicle-extra-candidates icicle-file-extras=
) (icicle-transform-function (quote icicle-remove-dups-if-extras)) (icicle-=
-temp-orders (copy-sequence icicle-sort-orders-alist)) (icicle-sort-orders-=
alist (progn (if (and icicle-file-sort-first-time-p icicle-file-sort) (prog=
n (setq icicle-sort-comparer icicle-file-sort icicle-file-sort-first-time-p=
 nil))) (if icicle-file-sort (let ((already-there ...)) (if already-there (=
cons already-there ...) (cons ... icicle--temp-orders))) icicle--temp-order=
s))) (icicle-candidate-help-fn (function (lambda (cand) (icicle-describe-fi=
le cand current-prefix-arg t)))) (icicle-candidate-alt-action-fn (or icicle=
-candidate-alt-action-fn (icicle-alt-act-fn-for-type "file"))) (icicle-all-=
candidates-list-alt-action-fn (or icicle-all-candidates-list-alt-action-fn =
(icicle-alt-act-fn-for-type "file"))) (icicle-delete-candidate-object (quot=
e icicle-delete-file-or-directory)) (icicle-candidate-action-fn (function (=
lambda (candidate) (let ((minibuffer-completion-table minibuffer-completion=
-table) (minibuffer-completion-predicate minibuffer-completion-predicate) (=
minibuffer-completion-confirm minibuffer-completion-confirm) (minibuffer-co=
mpleting-file-name minibuffer-completing-file-name) (minibuffer-completing-=
symbol ...) (minibuffer-exit-hook minibuffer-exit-hook) (minibuffer-help-fo=
rm minibuffer-help-form) (minibuffer-history-variable minibuffer-history-va=
riable) (minibuffer-history-case-insensitive-variables minibuffer-history-c=
ase-insensitive-variables) (minibuffer-history-sexp-flag minibuffer-history=
-sexp-flag) (minibuffer-message-timeout ...) (minibuffer-prompt-properties =
...) (minibuffer-setup-hook minibuffer-setup-hook) (minibuffer-text-before-=
history minibuffer-text-before-history)) (setq candidate (expand-file-name =
candidate ...)) (let (...) (cond ... ... ... ...)) (select-window (minibuff=
er-window)) (select-frame-set-input-focus (selected-frame)) nil))))) (icicl=
e-bind-file-candidate-keys) (let ((body (function (lambda nil (let (...) (s=
etq icicle-candidate-action-fn nil) (funcall ... file-choice)))))) (cond ((=
and debug-on-error debug-on-quit) (condition-case act-on-choice (funcall bo=
dy))) (debug-on-error (condition-case act-on-choice (funcall body) (quit (i=
cicle-try-switch-buffer icicle-orig-buff) nil))) (debug-on-quit (condition-=
case act-on-choice (funcall body) (error (icicle-try-switch-buffer icicle-o=
rig-buff) nil (error "%s" (error-message-string act-on-choice))))) (t (cond=
ition-case act-on-choice (funcall body) (quit (icicle-try-switch-buffer ici=
cle-orig-buff) nil) (error (icicle-try-switch-buffer icicle-orig-buff) nil =
(error "%s" (error-message-string act-on-choice))))))) (prog1 (setq dir-nam=
es (nreverse (delete "" dir-names))) (icicle-unbind-file-candidate-keys) (i=
f (interactive-p) (progn (message "Directories: %S" dir-names)))))
  icicle-directory-list()
  call-interactively(icicle-directory-list record-it)
  icicle-execute-extended-command-1("icicle-directory-list")
  #[...] 9]()
  funcall(#[...] 9])
  icicle-execute-extended-command()
  call-interactively(icicle-execute-extended-command nil nil)
 
In GNU Emacs 24.2.50.1 (i386-mingw-nt5.1.2600)
 of 2012-10-15 on DANI-PC
Bzr revision: 110553 monnier@iro.umontreal.ca-20121015164957-6zms5w2js1xkld=
tg
Windowing system distributor `Microsoft Corp.', version 5.1.2600
Configured using:
 `configure --with-gcc (4.7) --no-opt --enable-checking --cflags
 -I../../libs/libxpm-3.5.8/include -I../../libs/libxpm-3.5.8/src
 -I../../libs/libpng-1.4.10 -I../../libs/zlib-1.2.6
 -I../../libs/giflib-4.1.4-1/include -I../../libs/jpeg-6b-4/include
 -I../../libs/tiff-3.8.2-1/include
 -I../../libs/libxml2-2.7.8-w32-bin/include/libxml2
 -I../../libs/gnutls-3.0.16/include
 -I../../libs/libiconv-1.14-2-mingw32-dev/include'
 

bug#12677: 24.2.50; `nreverse' complains that ("foo") is not `listp'

[Stefan Monnier]

> Debugger entered--Lisp error: (wrong-type-argument listp ("c:/drews-lisp-20=
> /CONTRIB/"))

My guess is that the code ended up doing something morally equivalent
to (nreverse '(1 2 . 3)).  Try it and you'll see a similarly confusing
error message.
The reason is that by the time you get to the error, the list has
already been "nreversed" by side-effect, so the arg (which still points
to the cons cell which used to be the first element of the line) now
still points to the same cons-cell, but that cons-cell's cdr has been
set to nil because it was to become the last element of the
returned list.

I've installed a change in trunk which makes the above error include the
problematic non-nil tail, rather than the first element who looks
confusingly normal.


        Stefan

bug#12677: 24.2.50; `nreverse' complains that ("foo") is not `listp'

[Drew Adams]

> My guess is that the code ended up doing something morally equivalent
> to (nreverse '(1 2 . 3)).  Try it and you'll see a similarly confusing
> error message.

I understand your guess, but I think that might not be what was happening.  In
the debugger I evaluated the variable and saw that its full value was just that
singleton list.

But I'm OK with your closing this, as I cannot repro it from emacs -Q and it
does not interfere with my code.

> The reason is that by the time you get to the error, the list has
> already been "nreversed" by side-effect, so the arg (which 
> still points to the cons cell which used to be the first element
> of the line) now still points to the same cons-cell, but that
> cons-cell's cdr has been set to nil because it was to become the
> last element of the returned list.

Do I understand correctly that a true list with a single element should not
present such a problem?

> I've installed a change in trunk which makes the above error 
> include the problematic non-nil tail, rather than the first
> element who looks confusingly normal.

Sounds good.

bug#12677: 24.2.50; `nreverse' complains that ("foo") is not `listp'

[Stefan Monnier]

>> My guess is that the code ended up doing something morally equivalent
>> to (nreverse '(1 2 . 3)).  Try it and you'll see a similarly confusing
>> error message.

> I understand your guess, but I think that might not be what was
> happening.  In the debugger I evaluated the variable and saw that its
> full value was just that singleton list.

Of course that's what you'd see because that's indeed the value *at that
point*, even though it had a different value before nreverse started
modifying the list.  By the time you get the error, the problem data
is gone.


        Stefan

bug#12677: 24.2.50; `nreverse' complains that ("foo") is not `listp'

[Drew Adams]

> Of course that's what you'd see because that's indeed the 
> value *at that point*, even though it had a different value
> before nreverse started modifying the list.  By the time
> you get the error, the problem data is gone.

Got it.  Thx.