Re: Refile error - Kill is not a (set of) trees

[Chris Mann <cshclm@gmail.com>]

Carsten Dominik <carsten.dominik@gmail.com> writes:

> could you please do the following:
>
> - Run Emacs with a minimal configuration
> - Make me a test file that will let me reproduce the bug if possible.


Hello Carsten,

I've also been experiencing this behaviour lately.  I have been able to
reproduce this consistently with the following:

1) Create a file `reproduce.org', with the following contents:

   * Level1-1
   ** Level2-1
   *** TODO Level3-1
       SCHEDULED: <2009-08-05 Wed>
   ** Level2-2

2) run `emacs -Q' and evalute something resembling the following:

   (progn
     (add-to-list 'load-path "path/to/org-mode/lisp") 
     (require 'org-agenda)
     (setq org-agenda-files '("reproduce.org") 
           org-refile-targets '((("reproduce.org") . (:maxlevel . 2)))))

3) 'M-x org-agenda-list'

4) Move point to the `Level3-1' item in the agenda, 'C-c C-w' and
   attempt to refile it to, for example, Level2-1.

5) Error: org-paste-subtree: The kill is not a (set of) tree(s) [...]


As far as I can tell, the error occurs only when the heading's source
buffer is folded at the `Level1-1' heading; I could not reproduce the
error while the outermost heading was unfolded, regardless of the level
of the refile source / target.

I've investigated this error briefly: I believe `org-back-to-heading'
moves backward (passed the correct heading), to the outer-most heading,
causing `org-end-of-subtree' to then search for the start of heading of
the wrong level (too shallow).

Resultingly, the killed region may extend passed the end of the target
subtree.  In the above example, I believe the killed text would include
both the level 3 heading and the subsequent level 2 heading.

My suspicion is that this is stemming from a problem within outline-mode.

org-mode 6.29a
GNU Emacs 23.1.50.1

 -- Chris