Basic visibility cycling bug in 8.3.6

Basic visibility cycling bug in 8.3.6

[Marcus Zibrowius]

[-- Attachment #1: Type: text/plain, Size: 956 bytes --]

I am running org-mode 8.3.6 (8.3.6-6-g194e51-elpa
@ .../.emacs.d/elpa/org-20161003/) in GNU Emacs 24.3.1
(x86_64-pc-linux-gnu, GTK+ Version 3.10.7).

In this version, visibility cycling does not work as expected.  Consider
the following three-line example file:

* a
** b
** c

Consider the following sequence of commands:
1.  Place cursor on the letter b.  
2.  In the menu, choose Org > Refresh/Reload > Refresh setup current
buffer.
     (I do not know how to access this command directly).  
3.  Press TAB on your keyboard.  

If I do this in org-mode 8.3.6, my buffer looks as follows:

* a
** b
...

In this minimalistic example, pressing TAB once more will bring back the
line with the c, but in more complex examples, some parts of the buffer
seem to "get lost".  Moreover, I often find my files garbled like this
when I open them from the agenda. 

If I apply the above sequence of commands using my built-in org-mode
7.9.3, everything is fine.



[-- Attachment #2: Type: text/html, Size: 1360 bytes --]

Re: Basic visibility cycling bug in 8.3.6

[Nicolas Goaziou]

Hello,

Marcus Zibrowius <marcus.zibrowius@hhu.de> writes:

> In this version, visibility cycling does not work as expected.  Consider
> the following three-line example file:
>
> * a
> ** b
> ** c
>
> Consider the following sequence of commands:
> 1.  Place cursor on the letter b.  
> 2.  In the menu, choose Org > Refresh/Reload > Refresh setup current
> buffer.

Note that you are leaving, on purpose, point in an invisible part of the
document.

>      (I do not know how to access this command directly).  
> 3.  Press TAB on your keyboard.  
>
> If I do this in org-mode 8.3.6, my buffer looks as follows:
>
> * a
> ** b
> ...
>
> In this minimalistic example, pressing TAB once more will bring back the
> line with the c, but in more complex examples, some parts of the buffer
> seem to "get lost".  Moreover, I often find my files garbled like this
> when I open them from the agenda.

You can reveal text around point with C-c C-r (or even C-u C-c C-r). See
`org-reveal' for details.

By the way, Org 9.0 (yet to be released) contains a variable to control
this: `org-show-context-detail'. You may want to update Org.

> If I apply the above sequence of commands using my built-in org-mode
> 7.9.3, everything is fine.

I can reproduce the same with Org 7.9.3.

Regards,

-- 
Nicolas Goaziou

Re: Basic visibility cycling bug in 8.3.6

[Nicolas Goaziou]

Hello,

Marcus Zibrowius <marcus.zibrowius@hhu.de> writes:

Cc'ing to the list for further discussion.

> Excellent!   Thanks a lot for helping so quickly.  I still maintain that
> the behaviour is different for me in org-mode 7.9.3, but I understand
> now that it's a (new?) feature rather than a bug.

I misread your version string. The behaviour you describe doesn't happen
in Org 7.9.3f, but was introduced in
2efbd0f138180fe14773d5485ad3c59332d6ddb3, i.e., between Org 7.9.3f and
and Org 7.9.4. I can indeed observe the same with the latter.

However this change never made it into the following release, i.e., Org
8.0. You're observing a very short-lived feature.

> Let me reiterate what I think is different , now that I understand
> better what is happening:  in org-mode 8.3.6, if I start with 
>
> * a
> ** b
> ** c
> ** d
> ** e
> ** f
> ** g
> ** h
>
> with point at e, then if I collapse with Shift+TAB and reexpand with
> TAB, I get only:
>
> * a...
> ** e
> ...
>
> with point at the beginning of the line **e, and I can reexpand the
> whole tree with C-c C-r.   So the line in which point was is somehow
> remembered.  In contrast, Shift+TAB followed by TAB in org-mode 7.9.3f
> (details below) brings back the whole tree directly, but forgets where
> point was:  point ends up just right of a.

I think it is important to preserve point when cycling view. It would be
surprising if a function modifying visibility also moved it. Besides,
this is useful when one wants to have a glimpse to the overview (S-TAB)
and go back to the editing location (another S-TAB) immediatly after.

I have no strong opinion on what should be displayed when pressing TAB
in this case, tho. It seems the implentation of `org-cycle' sort of
decided for us, i.e., it wasn't meant to handle this situation.

Regards,

-- 
Nicolas Goaziou                                                0x80A93738