Re: [O] feature request: reloading settings in an org file should reload file local variables

[Carsten Dominik]

On 14 Jan 2014, at 16:33, Nick Dokos <address@hidden> wrote:

> Carsten Dominik <address@hidden> writes:
> 
>> Hi Eric,
>> 
>> On Jan 14, 2014, at 3:03 PM, Eric S Fraga <address@hidden> wrote:
>> 
>>> Hello,
>>> 
>>> I constantly run into an annoying side effect of org's handling of
>>> settings in a document.  When writing a document for export, usually to
>>> LaTeX, I find myself playing around with the document wide
>>> setting variables, e.g. 
>>> 
>>>  #+options: toc:nil
>>>  #+latex_header: \usepackage{tikz}
>>> 
>>> and similar.  Org requires me to re-load the document settings by
>>> hitting C-c C-c on any such document setting line.  This is fine.
>>> 
>>> The problem is that org appears to reset *all* variables to default
>>> (custom?)  settings.  This includes variables that have been set using
>>> file local variables in the given document.  Of particular note is the
>>> variable ~org-export-allow-bind-keywords~.  I normally have this set to
>>> nil for safety reasons but wish to set this to t in some documents.  I
>>> use something like this quite often on documents I *know* are safe:
>>> 
>>> ,----
>>> | # Local Variables:
>>> | # org-confirm-babel-evaluate: nil
>>> | # org-export-allow-bind-keywords: t
>>> | # End:
>>> `----
>>> 
>>> The problem is that hitting C-c C-c on a settings line clears these
>>> variables.  Having the bind one reset is particularly confusing at
>>> times.  In the end, I have to kill the buffer and re-visit the file to
>>> have things set up properly.
>>> 
>>> Is it possible to have org reload the whole file when reloading
>>> settings?  Or not reset any variables at all?  Either approach would be
>>> exhibit more consistent behaviour possibly?  Obviously, the position in
>>> the file would have to be remembered should the file be reloaded
>>> automatically.
>>> 
>>> This is not a major issue but a niggly one...
>> 
>> Could you try the following patch?
>> 
>> - Carsten
>> 
>> diff --git a/lisp/org.el b/lisp/org.el
>> index 549f1f8..6fca255 100644
>> --- a/lisp/org.el
>> +++ b/lisp/org.el
>> @@ -20482,11 +20482,9 @@ This command does many different things, depending 
>> on context:
>> Also updates the keyword regular expressions."
>>   (interactive)
>>   ;; this will set the mode *and* set file local variables.
>> -  (normal-mode)
>> -  ;; but it may leave us in some unrelated mode
>> -  (unless (derived-mode-p "org-mode")
>> -    (org-mode))
>> -  (message "Org-mode restarted"))
>> +  (funcall major-mode)
>> +  (hack-local-variables)
>> +  (message "Mode restarted"))
> 
> 
> I thought this was fixed, originally with commit 5ea0228
> and after York Zhao found a problem (see the thread at
> http://thread.gmane.org/gmane.emacs.orgmode/79187)
> with commit 0911edfac881ab17a2cd1670f169ff4bdf1ac486
> and two subsequent commits by Achim to fix some problems:
> 
> commit e655e664bff9fb6e98478682f03e713f990acba5
> commit f2483ec4bbe196ab29ef6312b0d7c1c179a9eafe
> 
> Is this another maint vs master issue perhaps?
> 
> Nick
> 
> 


Indeed, there were differences between maint and master.  Now fixed in both.
- Carsten