Re: Statistic cookies for headings and list items

[Bastien <bzg@gnu.org>]

Hi Michael,

thanks for reporting this.

Michael Brand <michael.ch.brand@gmail.com> writes:

> Is this all intended behaviour?

Well, no, I think the current behavior is confusing.

> When I start with ~C-c C-c~ on [ of line A, Org seems to count list items:
> Then ~S-<left>~ on line D seems to count subheadings:
> Then ~C-c C-c~ on [ of line A seems to count list items again:
> Then ~C-c -~ on line D makes D a subitem which makes no sense to me:
> But when I start with this:
> Then ~C-c -~ on line D makes D a sibling which I prefer to the above:
> Except that the automatic update like ~C-c C-c~ on [ of line A is missing:

I believe we cannot fix this without a discussion on the design first.

Here are a few solutions I can imagine:

1. when an entry contains both a list (as its direct contents) and
   subheadings, only consider subheadings in the stats calculation.

2. when an entry contains both a list (as its direct contents) and
   subheadings, only consider the list in the calculation.

3. if one of the two options above, allow the user to use a custom
   property to change the default (e.g. CUSTOM_STATS: list/headings)
   and consider the list of the subheadings.

4. add a new syntax rule to consider that stats at the beginning of 
   a headline are always for subheadings, while stats at the end of 
   a headline are always for the first list in direct contents.

I'd be in favor of (1) (without (3)) to keep things simple, but 
maybe that's a good opportunity to consider (4).  I think (3) is
only relevant if we go for (2), which I don't really like.

What do you think?

-- 
 Bastien