Re: [Orgmode] capture initial "level" and refile of capture buffer

[Carsten Dominik]

Hi Noorul, hi Richard,


On Oct 23, 2010, at 12:16 PM, Noorul Islam wrote:

> On Sat, Oct 23, 2010 at 2:12 PM, Noorul Islam <address@hidden>  
> wrote:
> > On Fri, Oct 22, 2010 at 5:13 PM, Richard Riley <address@hidden 
> > > wrote:
> > > What determines the level of a new capture element? e.g I just  
> > > created
> > > one and it started at "****".
> > >
> > > feature request : when I added some sub elements to a capture  
> > > buffer e.g
> > >
> > > * my new capture
> > >
> > > ** sub point
> > >
> > > *** sub sub point 1
> > > *** sub sub point 2
> > >
> > > and hit C-c C-w to refile, it only refiled the sub element (where  
> > > cursor
> > > was) and then lost the rest. I would like to suggest that refile  
> > > from
> > > the capture buffer should refile the entire buffer and not only the
> > > "current nested org item". Or am I missing something in my setup?
> >
> > On my box I have this observation.
> >
> > If I have something like this in my capture buffer
> >
> > * TODO Test
> >
> > * my new capture
> >
> > ** sub point
> >
> > *** sub sub point 1
> > *** sub sub point 2
> >
> >
> > and if I press C-c C-w at the last line (*** sub sub point 2) and
> > refile it to refile.org then what I get in refile.org is this
> >
> >
> > * TODO Test
> >
> > * my new capture
> >
> > ** sub point
> >
> > *** sub sub point 1
> > * sub sub point 2
> >
> >
> > The last one's level got changed.
> > I have latest pull from git repo.
> >
> > Org-mode version 7.01trans (release_7.01h.833.g21ad0)
> > GNU Emacs 23.2.2 (i686-pc-linux-gnu)
> >  of 2010-06-08 on sajida
>
> Fix org-capture bug.
>
> * lisp/org.el (org-capture-refile): Consider entire temporary buffer
> for refiling.


> diff --git a/lisp/org-capture.el b/lisp/org-capture.el
> index 7915f7f..6c62114 100644
> --- a/lisp/org-capture.el
> +++ b/lisp/org-capture.el
> @@ -548,7 +548,7 @@ already gone."
>    (unless (eq (org-capture-get :type 'local) 'entry)
>      (error
>       "Refiling from a capture buffer makes only sense for `entry'- 
> type templates"))
> -  (let ((pos (point))
> +  (let ((pos (point-min))
>         (base (buffer-base-buffer (current-buffer)))
>         (org-refile-for-capture t))
>      (org-capture-finalize)


This patch catures the problem and brings forward a correct idea.
But I believe we need to think a bit further.  The current
implementation of C-c C-w as a way to finish capture works
as closely as possible to the standard refile mechanism.
I.e. the tree *where the cursor is at* will be refiled.

Your patch makes the cursor move back to beginning of the
accessible part of the buffer and refiles from there.

If the captured entry starts at this point, and if all
the narrowed section contains is a single tree, this
is a good solution. However, if we move away from having
refile work in the exact same way as normally, the
following questions arise:

1. Maybe the user has entered an empty line before the subtree.
   In this case, the outline node *before* the captured tree
   will now be refiled.

2. Maybe the user has widened the capture buffer.  In this case
   the code will now refile the first node in the buffer, possibly
   very far from the current location of point.  Or, it will
   throw an error because the may not be an outline
   node at point min.

3. Maybe the user has added several trees (siblings) into the
   capture buffer.  In this case, the refile will only
   affect the first of those siblings.

So we have two solutions here:

Solution 1:  Be aware that you are just calling refile, so
             move the cursor to the appropriate place before
             running the command.

Solution 2:  Install two markers at the beginning and end of
             the region to which the buffer has been narrowed,
             and then select that entire region for refiling.
             This is non-trivial, but probably can be made to
             work.

In either case we will probably have to update the manual
to make things clearer.

- Carsten