[Orgmode] Re: moving in the agenda view is slow

[Nick Dokos]

Eric S Fraga <address@hidden> wrote:

> I generate results for a number (looks like 99 times from the results)

Looks like 129 times from the results below.

> of 'n' moves through my agenda, including going through a few days from
> today forwards.  I get the following:
> 
> ,----
> | org-agenda-later                     4           2.319228      0.579807
> | org-agenda-redo                      4           2.319001      0.57975025
> | org-agenda-list                      5           1.907568      0.3815136
> | org-agenda-get-day-entries           99          1.7082340000  0.0172548888
> | org-agenda-to-appt                   4           1.170569      0.29264225
> | org-agenda                           1           1.161838      1.161838
> | org-let                              4           1.145839      0.28645975
> | org-agenda-get-scheduled             99          0.8759350000  0.0088478282
> | org-prepare-agenda                   5           0.8665130000  0.1733026000
> | org-prepare-agenda-buffers           9           0.721267      0.0801407777
> | org-agenda-next-line                 129         0.6453539999  0.0050027441
> `----
> 

Are you leaning on the "n" key? It's probably better to press it a given
number of times instead. Or is it the case that the delay you and Rainer
see *only* exhibits itself on auto-repeat? If the latter, then it might
very well be the case that X is the culprit (or the emacs display engine
or who knows what else).

The org-agenda-next-line time per call is about 4x what I get on my
system (where I have not seen the problem), so I assume that you will
also probably get the 50x (or so) slowdown on your office system that
Rainer got.

> Now, this is on my home system on which the response is actually
> perfectly fine!  It is on my work system where the response is slow so I
> will repeat this on Tuesday if I get a chance.  My home machine is
> actually slower than my office system *but* my office system is using a
> less effective X window system graphics driver so the current view that
> font-locking may have something to do with the problem could be
> consistent with this.
> 

Actually, I'm not sure it has anything to do with font-lock: I got some
font-lock results in my initial profile (for unknown reasons - I hadn't
added font-lock to the elp list afaik, but I may have done something
stupid), and Manuel Hermenegildo chimed in with a problem that he has
had for a while: there is a suspected font-lock connection there and
Manuel posted a workaround for this problem[fn:1], but afaik this
problem is not related to that one.

The main thing to check in this problem case is whether the time that
org-agenda-next-line takes is roughly the same as the time that
next-line itself takes. There is somewhat indirect evidence that that is
the case, but we need to make sure. I asked Rainer to add next-line to
the elp list with M-x elp-instrument-function <RET> next-line <RET>, but
I haven't heard any results yet.

Thanks,
Nick

Footnotes:
[fn:1] I don't remember who posted the workaround originally and haven't
gone back to check.