bug#15418: 24.3.50; RCS annotate broken in vc-rcs.el r114131

[Richard Copley]

The top of the stack trace is exactly the same as before, except for the new name for "b".

  buffer-substring-no-properties(nil 6747)
  (cons (buffer-substring-no-properties (cadr pairs) (car pairs)) acc)
  (setq acc (cons (buffer-substring-no-properties (cadr pairs) (car pairs)) acc))
  (while pairs (setq acc (cons (buffer-substring-no-properties (cadr pairs) (car pairs)) acc)) (setq pairs (cddr pairs)))
  (let ((pairs (cons e (append @-holes (list vc-rcs--b)))) acc) (while pairs (setq acc (cons (buffer-substring-no-properties (cadr pairs) (car pairs)) acc)) (setq pairs (cddr pairs))) (apply (quote concat) acc))

(The next line of the backtrace has 159030 characters, most of which are from the master file. Sorry again that I can't provide an example, but the file doesn't belong to me.)

On 25 September 2013 01:55, Stefan Monnier <monnier@iro.umontreal.ca> wrote:

>> > For some files under RCS version control, `vc-annotate' now fails with
>> > error `(wrong-type-argument integer-or-marker-p nil)'.
>> > I think the problem is with the variable `b' whose dynamic binding
>> > was used by `gather' and set by `incg' (local functions in
>> > `vc-rcs-parse'), prior to r114131.
>> Indeed, it seems like there was some subtle dynamic scoping at play.
>> Does the patch below fix your problem?
> No, I'm afraid it does not. I get the same error.

Can you show the new backtrace?


        Stefan