Re: [PATCH] Optimize org-habit-parse-todo

[Matt Lundin <mdl@imapmail.org>]

* lisp/org-habit.el: (org-habit-parse-todo) Don't parse more days than
  needed.

When constructing a consistency graph, org-habit now stops searching
for timestamps when the number of matches exceeds the span of time
displayed in the graph. This can lead to a significant speedup in
agenda construction, especially for entries with many logbook entries.
Previously, org-habit would parse all logbook timestamps, even if they
numbered in the hundreds.
---
 lisp/org-habit.el |   16 ++++++++++++----
 1 files changed, 12 insertions(+), 4 deletions(-)

diff --git a/lisp/org-habit.el b/lisp/org-habit.el
index b174a1f..5d2514a 100644
--- a/lisp/org-habit.el
+++ b/lisp/org-habit.el
@@ -170,10 +170,18 @@ This list represents a \"habit\" for the rest of this module."
 		   habit-entry scheduled-repeat))
 	(setq deadline (+ scheduled (- dr-days sr-days))))
       (org-back-to-heading t)
-      (while (re-search-forward "- State \"DONE\".*\\[\\([^]]+\\)\\]" end t)
-	(push (time-to-days
-	       (org-time-string-to-time (match-string-no-properties 1)))
-	      closed-dates))
+      (let* ((maxdays (+ org-habit-preceding-days org-habit-following-days))
+	     (reversed org-log-states-order-reversed)
+	     (search (if reversed 're-search-forward 're-search-backward))
+	     (limit (if reversed end (point)))
+	     (count 0))
+	(unless reversed (goto-char end))
+	(while (and (< count maxdays)
+		    (funcall search "- State \"DONE\".*\\[\\([^]]+\\)\\]" limit t))
+	  (push (time-to-days
+		 (org-time-string-to-time (match-string-no-properties 1)))
+		closed-dates)
+	  (setq count (1+ count))))
       (list scheduled sr-days deadline dr-days closed-dates))))
 
 (defsubst org-habit-scheduled (habit)
-- 
1.7.3.5

Carsten Dominik <carsten.dominik@gmail.com> writes:

> On Jan 25, 2011, at 7:01 AM, Carsten Dominik wrote:
>
>> Hi Matt
>>
>> Hmmm,
>>
>> this looks like a very important optimisation indeed.
>> I am just wondering if it is always safe to do it like
>> this.  Have you checked if this is influenced by
>> org-reverse-notes-order or similar things?
>
> I am sorry, I see now that this is done correctly.
> One request, can you resubmit and test for the count
> first, before doing the search?  Just another very
> minor optimization.

Thanks Carsten!

See the updated patch above. The next step is to make the keyword search
configurable....

Best,
Matt