[Orgmode] Re: Strange bug, request for more info

[Carsten Dominik]

On May 26, 2010, at 4:44 AM, Bernt Hansen wrote:

> Carsten Dominik <address@hidden> writes:
>
> > Hi everyone,
> >
> > John Wiegley is being haunted by a strange bug and I have so far not
> > been able to reproduce and fix it.  So I would like to know
> > if anyone else sees the same bug and can contribute observations
> > that may help us to track this down.  I believe I had a similar
> > report quite a while ago, but don't remember who reported it.
> >
> > The bug happens when being in the agenda and trying to goto or show
> > the origin location of an agenda entry by pressing SPC or RET.
> > John reports that sometimes (for him several times a day),
> > the other window shows a completely different location.
> > The most weird part of it is that going back to the agenda buffer
> > and then trying the exact same command again, everything works
> > fine!  This is driving me crazy, and I'd love to find and fix
> > this problem.
> >
> > So please, if anyone sees the same bug, try to give as as much as  
> > info
> > as possible.  How often does it happen, under what circumstances,
> > what is your setup etc etc.
>
> Hi Carsten,
>
> Here I am replying to a 2.5 year old post...  I assume this hasn't  
> been
> fixed yet.  I think I was the other person who mentioned this ages  
> ago.
> I'm still seeing this 'go to the wrong place' scenario up to a few  
> times
> per day.
>
> It's really annoying - you 'go to' a task and it's the wrong place --
> but if you do it again it works.  I've seen this occur in the  
> following
> cases:
>
> - looking at an agenda and hitting SPC or TAB to get to the task
> - jumping to the last refile task with C-u C-u C-c C-w
> - jumping to the last remember task with C-u C-u C-M-r
>
> With follow mode on I've never noticed this bug but I don't use
> follow-mode regularly.
>
> Let me know if you can think of anything I can provide to help track
> down this bug.

Unfortunately, I really don't know how to fix this bug.   Until I can  
get a way to reproduce this, it will be very hard to figure out what  
is causing this.

My suspicion is that it has to do with redisplay.  When Emacs is  
redisplaying, it will move window start in a window so that (point)  
will be visible in the buffer.  For some reason that is a complex  
combination of folding text (maybe) and the complex code that Org  
evaluates before actually showing an agenda item, it seems that the  
redisplay engine might be kicked of with a wrong idea of where point  
should be.

In this view, the reason why Samuel is getting this error more than  
anybody else would have to do with the fact that he is using HUGE  
fonts, with very little text in a window at any given time, so that  
the "point is not in the window" condition might be triggered more  
often.

These are wild guesses, I know that years ago I spent *many* hours  
trying to figure out what is happening her.  I still don't know.

What you can do is make me a test case that will reproduce this  
behavior, reliably.  I think that this problem could depend on hook  
settings and on setings of the context-revealing variables like

org-show-hierarchy-above
org-show-following-heading
org-show-entry-below
org-show-siblings

so if you are trying to make a test case, you should use the same  
values for these variables and also try to figure out if the current  
visibility state of the outline has to do with it.

Also, does this error exist in Emacs 23?

:/

Regards

- Carsten