Re: [BUG] `org-fill-paragraph' doesn't use `fill-prefix'

[Daniel Hackney <dan@haxney.org>]

Nicolas Goaziou <n.goaziou@gmail.com> wrote:
> Daniel Hackney <dan@haxney.org> writes:
>
>> I proposed a fix [1] for this back in 2010, but it seems to have regressed
>> again. `org-fill-paragraph' no longer makes use of a `fill-prefix', so
>> filling things like email comments no longer works.
>
> It has been discussed on this ML already. Org mode is not Message mode
> and ">" prefix means nothing to it.

It isn't just ">" that behaves this way in `fundamental-mode' and older
versions of org-mode; there are a number of such characters. From a
quick test, in Emacs 24.2,

#+BEGIN_EXAMPLE
-- foo
-- bar
#+END_EXAMPLE

becomes

#+BEGIN_EXAMPLE
-- foo bar
#+END_EXAMPLE

whereas in the current org-mode, it is

#+BEGIN_EXAMPLE
-- foo -- bar
#+END_EXAMPLE

Same with

#+BEGIN_EXAMPLE
! foo
! bar
#+END_EXAMPLE

and

#+BEGIN_EXAMPLE
% foo
% bar
#+END_EXAMPLE

The following also fill to a single line:

#+BEGIN_EXAMPLE
% % foo
% % bar

! ! foo
! ! bar
#+END_EXAMPLE

From a brief look at `fill-paragraph', the relevant code is in
`fill-comment-paragraph'. I might take a closer look at it to see if I
can come up with my own fix.

> Also, it has its own set of special prefixes, which are not found in
> Fundamental mode. For example, you can never have " : " as a fill prefix
> since it creates a fixed-width area.

Of course. Fundamental mode will fill

#+BEGIN_EXAMPLE
- foo
- bar
#+END_EXAMPLE

into a single line, but org-mode never did. I'm not asking for new
behavior, simply the `fill-paragraph' style of org 7.8.11 back.

> Therefore, I don't consider it to be a regression since it's not an
> expected feature in the first place. But I admit it is still
> convenient.

Well, my patch from 2010 /was/ applied [1], so I would consider it an
expected feature :)

> Maybe we can introduce some support for `adaptive-fill-regexp' in
> paragraphs and comments filling. Would you mind testing the following
> patch?

I'll test it later today.

[1] http://article.gmane.org/gmane.emacs.orgmode/22954

--
Daniel Hackney