[O] In agenda, better handling of files changed behind Org's back (was: Agenda highlighting bug)

2012-12-29 Thread Bastien
Hi Simon,

(New thread for this topic.)

Simon Thum simon.t...@gmx.de writes:

 I'd also like to use this opportunity to ask about more graceful agenda
 handling of changed files. When someone (not emacs) touches agenda files,
 the agenda code fails horribly.   That is even if the file is readonly or has
 the #+READONLY header (which seems menaingful only to org-mobile). It
 would be preferable to just reload in this case. ATM, I am queried about
 something related to reloading and the agenda cancels no matter what I
 say.

I get the gist of what may happen here, but I'd need a reproducible 
recipe to know where to put the checks.  Sorry to put the burden on
your shoulders, but this would help a lot!

Thanks,

-- 
 Bastien



Re: [O] In agenda, better handling of files changed behind Org's back

2012-12-29 Thread Simon Thum

On 12/29/2012 11:29 AM, Bastien wrote:

Hi Simon,

(New thread for this topic.)

Simon Thumsimon.t...@gmx.de  writes:


I'd also like to use this opportunity to ask about more graceful agenda
handling of changed files. When someone (not emacs) touches agenda files,
the agenda code fails horribly.   That is even if the file is readonly or has
the #+READONLY header (which seems menaingful only to org-mobile). It
would be preferable to just reload in this case. ATM, I am queried about
something related to reloading and the agenda cancels no matter what I
say.


I get the gist of what may happen here, but I'd need a reproducible
recipe to know where to put the checks.  Sorry to put the burden on
your shoulders, but this would help a lot!


It's not a burden at all. This recipe breaks for me:

1 ) invoke agenda

2)  touch some_agenda_file.org

3) invoke agenda again

I am asked: some_agenda_file.org changed on disk; really edit the 
buffer? (y, n, r or C-h)


No matter what I answer, the agenda does not show up. Using r, I can get 
the agenda in a repeat call.


My argument is that read-only buffers are not supposed to change on 
emacs' side, so it would always be OK to re-read them.


Cheers,

Simon



Re: [O] In agenda, better handling of files changed behind Org's back

2012-12-29 Thread Bastien
Hi Simon,

Simon Thum simon.t...@gmx.de writes:

 It's not a burden at all. This recipe breaks for me:

 1 ) invoke agenda

 2)  touch some_agenda_file.org

 3) invoke agenda again

 I am asked: some_agenda_file.org changed on disk; really edit the buffer?
 (y, n, r or C-h)

 No matter what I answer, the agenda does not show up. Using r, I can get
 the agenda in a repeat call.

 My argument is that read-only buffers are not supposed to change on emacs'
 side, so it would always be OK to re-read them.

I understand and I agree.  

There is no simple fix, though.  I'll put this on my watch list.

Best,

-- 
 Bastien



Re: [O] In agenda, better handling of files changed behind Org's back

2012-12-29 Thread Simon Thum

On 12/29/2012 02:29 PM, Bastien wrote:

Hi Simon,

Simon Thumsimon.t...@gmx.de  writes:


It's not a burden at all. This recipe breaks for me:

1 ) invoke agenda

2)  touch some_agenda_file.org

3) invoke agenda again

I am asked: some_agenda_file.org changed on disk; really edit the buffer?
(y, n, r or C-h)

No matter what I answer, the agenda does not show up. Using r, I can get
the agenda in a repeat call.

My argument is that read-only buffers are not supposed to change on emacs'
side, so it would always be OK to re-read them.


I understand and I agree.

There is no simple fix, though.  I'll put this on my watch list.


Thanks for that. I assumed it would be easier.

I'm using a fair amount of generated org files, and what especially 
sucks with this behaviour is that _each changed file_ requires one 
agenda building attempt. So before re-reading 3 org files it is easier 
to restart emacs.


Cheers,

Simon



Re: [O] In agenda, better handling of files changed behind Org's back

2012-12-29 Thread Achim Gratz

Am 29.12.2012 14:29, schrieb Bastien:

My argument is that read-only buffers are not supposed to change on emacs'
side, so it would always be OK to re-read them.


If that is true (I don't really know), then auto-revert-mode should keep 
them in sync with the file on disk.



There is no simple fix, though.  I'll put this on my watch list.


I think that, aside from the above suggestion, it may be worth pondering 
if Org should really visit those files that it just needs to look at 
for building the agenda.  If it would not visit them (Emacs must visit a 
file that the user opened in edit mode obviously), the problem should 
not appear.



--
Achim.

(on the road :-)




Re: [O] In agenda, better handling of files changed behind Org's back

2012-12-29 Thread Bastien
Hi Achim,

Achim Gratz strom...@nexgo.de writes:

 I think that, aside from the above suggestion, it may be worth pondering if
 Org should really visit those files that it just needs to look at for
 building the agenda.  If it would not visit them (Emacs must visit a file
 that the user opened in edit mode obviously), the problem should not
 appear.

The problem is that Org does not only need to look at the files,
it also adds text properties for example, which is considered editing
by Emacs, even the modification only affects the buffer, not the file.

We could try to use `with-buffer-modified-unmodified' a bit more, but
I'm not convinved.

-- 
 Bastien