Re: M-RET inside the first second-level heading of the first first-level heading

[John Hendy <jw.hendy@gmail.com>]

On Tue, May 7, 2013 at 4:54 AM, Bastien <bzg@gnu.org> wrote:
> Hi John,
>
> John Hendy <jw.hendy@gmail.com> writes:
>
>> On Fri, Apr 26, 2013 at 1:54 AM, Bastien <bzg@gnu.org> wrote:
>>> Hi Jisang,
>>>
>>> Jisang Yoo <jisang.yoo.ac+org@gmail.com> writes:
>>>
>>>> 2. Place cursor at the beginning of "** bacon" and press M-RET and org creates
>>>> a first-level heading.
>>>
>>> This should now be fixed.  Thanks for reporting this,
>>
>> There were some same/similar/related issues as well, and I'm still not
>> getting perfect results.
>>
>> Using a minimal config (below), if I create this in a new file
>>
>> * test
>> ** test1
>> ** test2
>>
>> and then fold on * test, I get this:
>>
>> * test...2
>
> I do have (setq require-final-newline t) in my config, which prevents
> this.  What happens is this: if you don't have the config above, the
> folding will get wrong and display the "2" from the end of the buffer.
>
>> Also, after navigating to the end of * test...2 (with it folded) and
>> issuing M-RET, I get a new second level heading after ** test2. I
>> would have expected a new first level headline since I did M-RET on a
>> first level headline. Or is that the default behavior?
>
> If you are before the "..." ellipsis, you are on a first level
> headline and M-RET will insert a first level headline.  Otherwise you
> are on a second level headline ("test2") and it will insert a second
> level headline.

Ah, that now makes sense. But... just to be clear, take this case:

* Headline1
- list1
- list2

Now fold it:

* Headline1<cursor>...

If I do M-RET at <cursor>, I get:

* Headline1
* <cursor>
- list1
- list2

Is that how it's supposed to work? My use case is generally to have a
new headline after * Headline1 *and* it's contents, not putting
contents inside the new headline.

C-RET seems to behave more like I'd expect. In reading the manual, I
think I was just confused on usage based on the definition of C-RET,
however this in the description of M-RET is confusing:

#+begin_quote
If the command is used at the end of a folded subtree (i.e., behind
the ellipses at the end of a headline), then a headline like the
current one will be inserted after the end of the subtree. Calling
this command with C-u C-u will unconditionally respect the headline's
content and create a new item at the end of the parent subtree.
#+end_quote

It makes it seem like my original case in which the cursor is behind
(after?) the ... should insert a same-level headline after the end of
the current subtree (which I would assume means that headline and all
contents). Am I reading that incorrectly? It doesn't really talk about
the behavior if you're before the ellipsis.


Thanks for clarifying,
John

>
>> If so, I guess my only concern is the folding of the end of a headline
>> if there's no hard return after it into the ellipsis of it's parent.
>
> Yes.  Maybe M-RET could/should handle this corner-case but since
> `org-insert-heading' is deserving a full rewrite, I'm not going to
> try to handle this corner-case myself... let's just keep it in mind
> when doing the rewrite.  Thanks for spotting it,
>
> --
>  Bastien