bug#16247: syntax-ppss returns incorrect state

bug#16247: syntax-ppss returns incorrect state

[Barry OReilly]

[-- Attachment #1: Type: text/plain, Size: 1688 bytes --]

Let /tmp/get-syntax.el :

  (defun get-syntax ()
    (interactive)
    (message "%s" (syntax-ppss (point))))

Run Emacs from trunk:

  ./src/emacs -Q lisp/progmodes/cc-engine.el --load /tmp/get-syntax.el

Insert this in Emacs:

diff --git a/lisp/progmodes/cc-engine.el b/lisp/progmodes/cc-engine.el
index 973ed91..882c4f5 100644
--- a/lisp/progmodes/cc-engine.el
+++ b/lisp/progmodes/cc-engine.el
@@ -134,7 +134,7 @@
 ;;   cc-awk.el.

 ;;; Code:
-
+(
 (eval-when-compile
   (let ((load-path
         (if (and (boundp 'byte-compile-dest-file)

Go a couple lines down, | indicates point:

(
(eval-when-compile
|  (let ((load-path
        (if (and (boundp 'byte-compile-dest-file)

Execute M-x get-syntax and see:

  (2 5705 5706 nil nil nil 0 nil nil (5703 5705))

Looks ok. Move down in buffer, | indicates point:

;; Included in the mode line to indicate the active submodes.
;; (defvar c-submode-indicators nil)
;; (make-variable-buffer-local 'c-submode-indicators)
|
(defun c-calculate-state (arg prevstate)
  ;; Calculate the new state of PREVSTATE, t or nil, based on arg. If
  ;; arg is nil or zero, toggle the state. If arg is negative, turn
  ;; the state off, and if arg is positive, turn the state on
  (if (or (not arg)
      (zerop (setq arg (prefix-numeric-value arg))))
      (not prevstate)
    (> arg 0)))

M-x get-syntax:

  (0 nil 7000 nil nil nil 0 nil nil nil)

Expected the depth to be 1, not 0. For good measure, go back to the
same position of the first get-syntax call and do it again:

  (1 5705 5706 nil nil nil 0 nil nil (5705))

Why did the same buffer position change from depth 2 to depth 1, even
though the buffer didn't change between the get-syntax calls?

[-- Attachment #2: Type: text/html, Size: 1952 bytes --]

bug#16247: syntax-ppss returns incorrect state

[Stefan Monnier]

> (
> (eval-when-compile
> |  (let ((load-path
>         (if (and (boundp 'byte-compile-dest-file)
[...]
> Why did the same buffer position change from depth 2 to depth 1, even
> though the buffer didn't change between the get-syntax calls?

That's because of syntax-begin-function, which in Lisp buffers assumes
than an open-paren in column 0 is necessarily outside of anything
(parens, comments, strings, ...).  And then syntax-ppss sometimes uses
this assumption and sometimes not, depending on the distance of the
previous call to syntax-ppss and the nearest beginning of defun (aka
open-paren in column 0).

I just installed a change to leave syntax-begin-function nil in
Lisp buffers.  The performance impact should be negligible.


        Stefan