emacs-orgmode
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Orgmode] Re: Lazy evaluation when defining org-remember-template (aka E


From: Keith Swartz
Subject: [Orgmode] Re: Lazy evaluation when defining org-remember-template (aka Emacs-orgmode Digest, Vol 39, Issue 122)
Date: Mon, 01 Jun 2009 13:36:25 -0700
User-agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1.3) Gecko/20070326 Thunderbird/2.0.0.0 Mnenhy/0.7.5.666

Nick,

Just wanted to report that this solution worked perfectly. Thank you!

And thank you, also, to Robert for the suggested rewrite. Of course, as soon as I read your post, my hands froze above the keyboard for about two minutes while my brain exercised a massive page fault (it gets awfully fragmented in there after fifteen years), but it was worth it.

I'm inclined to agree with you, Nick, about making this change permanent. In this case, the most likely problem would be an error, and worst case scenario might be localized data loss that could probably be retrieved. (I can't conceive of what someone would enter to make that happen, though.) But still, I'm always very reluctant to have code perform an eval() on any user-entered data without intense scrubbing, and I'm fine making this one-line change to the code myself.

Thanks again,
Keith
Robert Goldman <address@hidden> wrote:

  
Date: Fri, 29 May 2009 23:24:58 -0700
From: Keith Swartz <address@hidden>
Subject: [Orgmode] Lazy evaluation when defining org-remember-template
To: "[orgmode]" <address@hidden>
Message-ID: <address@hidden>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

...

Is there a way I can make that command evaluate at the time it is 
invoked, rather than when it is defined? I vaguely recall doing 
something like this, but that was five job roles, three houses, two 
recessions, and two kids ago. :)

      
I can't swear that this will work, but note that the way you have
written this, it will all be evaluated at load time, as you say.  the
'list' function will evaluate its arguments to build the list.

Now, if you don't want this to be evaluated when org-remember-templates
is set, you can quote the form:

'(format-time-string "%A")

[note that you quoted the argument to format-time-string.  I don't
believe that's necessary, since strings evaluate to themselves, but I
have not tested this.]

Actually, I think you would get something easier to read if you quoted
the whole list, instead of quoting each element.  Something like:

(list '("Todo" ?t "* TODO %?%^{To do} %^g\n  :LOGBOOK:\n  -
Added: %U\n  :END:" "d:/tmp/_my.todo" (format-time-string "%A"))))
    
That's correct.
  
The question then is, "what happens when org-remember-templates is
retrieved?"  What you want is for this function to be evaluated when the
templates are found and used.  That will be done by
org-remember-apply-template, which we can examine....

Unfortunately, I don't see in there anything which retrieves (nth 4
entry), which is the place where your format-time-string goes, so I'm
not sure what is handling this.  It's a little confusing reading that
function's code, since "headline" is ambiguous between whether it means
the headline of the remember note to be inserted or the headline under
which to insert the note...  I believe it's the former.
    

It's the latter.

You can figure out things like this fairly quickly by inserting a
(debug) at the appropriate place, and re-evaluating the defun. When the
function gets called, it will jump into the debugger when it evals the
(debug) form, and you can use the full power of lisp to examine
state. For example, here I defined the template the way you suggested,
placed a (debug) in org-remember-apply-template, just after the
insertion of the template in the remember buffer, re-evaluated the defun
(there is an eval-defun, but I prefer to do that by going to the end of
the defun - which I can do quickly: repeat M-C-u until I'm at the
beginning of the defun and M-C-f to move over the whole defun - and then
C-x C-e to eval the last sexpression.)

I then call org-remember and in the resulting debug buffer, say

  e headline<RET>

which says 

(format-time-string "%A")

  e entry<RET>

which says

("* TODO %?%^{To do} %^g
  :LOGBOOK:
  - 
Added: %U
  :END:" (quote "d:/tmp/_my.todo") (format-time-string "%A"))

Now you can see that the headline is the third element of this list
(i.e. (nth 2 entry) - the numbering starts from 0).

  
Perhaps someone else can figure this out, or perhaps you could just try
quoting the list and seeing if it works to evaluate the
format-time-string when you want it to.  Org usually does The Right Thing.

    
But even org cannot perform miracles !-) Somebody has to "force the thunk"
in order for delayed evaluation to work. You'd need something like this 
patch:

--- a/lisp/org-remember.el
+++ b/lisp/org-remember.el
@@ -388,7 +388,7 @@ to be run from that hook to function properly."
 				(functionp (nth 1 entry))))
 		       (nth 1 entry)
 		     org-default-notes-file))
-	     (headline (nth 2 entry))
+	     (headline (eval (nth 2 entry)))
 	     (v-c (and (> (length kill-ring) 0) (current-kill 0)))
 	     (v-x (or (org-get-x-clipboard 'PRIMARY)
 		      (org-get-x-clipboard 'CLIPBOARD)

This should work in simple cases (in particular, because the headline is
a string and strings evaluate to themselves, so it should not adversely affect
any existing template), but I certainly have not thought about repercussions
(including the possibility of *very* obscure bugs because somebody mistyped
something in the template - that would be a maintenance nightmare that Carsten
might not be willing to take on).

Thanks,
Nick
  


reply via email to

[Prev in Thread] Current Thread [Next in Thread]