Re: Capture template and elisp expression

[Carsten Dominik <carsten.dominik@gmail.com>]

On Jan 7, 2011, at 1:08 PM, Štěpán Němec wrote:

> Carsten Dominik <carsten.dominik@gmail.com> writes:
>
>> On Jan 6, 2011, at 11:26 PM, Nick Dokos wrote:
>>
>>> Rainer M Krug <r.m.krug@gmail.com> wrote:
>>>
>>>>
>>>> Hi
>>>>
>>>> I want to use a Capture Template to record changes to files under
>>>> version control. Everything works as expected, but I would like to
>>>> include the current revision in the template.
>>>>
>>>> Therefore I tried the following:
>>>>
>>>> * %T %? at %a by Rainer M Krug, email: Rainer@krugs.de
>>>> %(vc-working-revision buffer-file-name)
>>>>
>>>> for the template, but I get an error:
>>>>
>>>> **** <2011-01-06 Thu 13:06>  at
>>>> [[file:~/Documents/Projects/BiocontrolAndAlienDynamics/
>>>> nonSpatialAcacia/trunc/R/nsa.org::*Finalise][Finalise]]
>>>> by Rainer M Krug, email: Rainer@krugs.de %![Error: (wrong-type-
>>>> argument
>>>> stringp nil)]
>>>>
>>>> Any idea how I could get the revision of the org file from which  
>>>> the
>>>> Capture has been initiated (here
>>>> ~/Documents/Projects/BiocontrolAndAlienDynamics/nonSpatialAcacia/
>>>> trunc/R/nsa.org
>>>> )?
>>>>
>>>
>>> There are a few problems: the evaluation of the sexp happens in the
>>> capture buffer where buffer-file-name returns nil. Even if you could
>>> get the file name, vc-working-revision would return nil on a file  
>>> that
>>> is not VC-registered and the template would barf.
>>>
>>> Those are easy problems to solve but there is another one that seems
>>> insurmountable (with current code): my original thought was to use  
>>> the
>>> %a escape to pass the link to a lisp function, extract the file name
>>> from it[fn:1] and run vc-working-revision on it (with appropriate  
>>> safeguards
>>> to catch non-VC files), something like this:
>>>
>>> "* %T %? at %a by Rainer M Krug, email: Rainer@krugs.de        %(rk-
>>> custom-function-to-get-vc-revision \"%a\")"
>>>
>>> However, this fails because at the time that %(sexp) constructs are
>>> expanded, simple %a etc. constructs have not been expanded yet, so  
>>> what
>>> the function above gets is a literal "%a": the subtitution  
>>> sequence is
>>>
>>>           ;; %[] Insert contents of a file.
>>>
>>>           ...
>>>
>>>           ;; %() embedded elisp
>>>
>>>           ...
>>>
>>>           ;; Simple %-escapes
>>>
>>> (see lisp/org-capture.el, lines 1181-1229 or so).
>>>
>>> Moreover, this sequence was different and was changed deliberately  
>>> (see
>>> the thread http://thread.gmane.org/gmane.emacs.orgmode/27649), so  
>>> if it
>>> is changed back, Sebastion Rose will not be happy :-)
>>>
>>> So it seems there is no way to pass values from the capture  
>>> context to a
>>> lisp function in the capture template, but maybe I'm missing  
>>> something.
>>>
>>> Thanks,
>>> Nick
>>>
>>> Footnotes:
>>> [fn:1] Is there an easier way to get the filename of the file I was
>>> visiting when I initiated the capture? If not, should there be?  
>>> Perhaps
>>> a %f escape?
>>
>> Hi Nick,
>>
>> you can use
>>
>>    (buffer-file-name (org-capture-get :original-buffer))
>>
>> and we could certainly introduce a special escape for it if helpful.
>>
>> If it is easier, we can also put the filename itself into the  
>> property list,
>> and any other information we like.  This should happen in the  
>> function
>> org-capture,
>> close to the location where the buffer is stored, so near this line:
>>
>> 	(org-capture-put :original-buffer orig-buf :annotation annotation
>> 			 :initial initial)
>>
>> org-capture uses this property list precisely so that it is simple
>> to add any information required.
>>
>> Note that, after the template has been filled in, it is better
>> to access information in the property list with
>>
>>
>>  (org-capture-get PROPERTY 'local)
>>
>> to avoid conflicts with other ongoing capture processes.
>>
>> Hope this helps.
>
> Why aren't the %() expressions simply evaluated in the original buffer
> (if available)? That would solve these issues in a general way. It  
> seems
> to me that there is no advantage to evaluating the expressions in the
> temporary capture buffer, but I'm not familiar with the code so I  
> might
> be missing something. Is there a reason for that?

The sexp can be used to insert stuff into the template, so I think it
is correct to evaluate it in the template buffer.

- Carsten