Massive Org search

Massive Org search

[François Pinard]

Hi, Org people.

I'm a rather recent Org mode user.  Consolidating my previous data into
Org mode gave 385 files using 5,5 Mb.  I quickly found out that this is
too much for Org mode, if I want to be able to search it all using Org
mode means.  So, I segregated a handful of agenda files, and developed a
small tool for when I need a massive org search (which is often).  Let
me share it, in case useful to others.  For the code, see:

http://stackoverflow.com/questions/6554110/emacs-outline-mode-auto-expand-upon-grep/8439425

The idea is simple: just use a recursive grep and normal compile-type
buffer and commands, yet add a bit of code to reveal hidden text which
is being located by the compile commands.  Also manage so that a mere
C-s on the located buffer immediately show all occurrences.

François

Re: Massive Org search

[Bastien]

Hi François,

pinard@iro.umontreal.ca (François Pinard) writes:

> I'm a rather recent Org mode user.  Consolidating my previous data into
> Org mode gave 385 files using 5,5 Mb.  I quickly found out that this is
> too much for Org mode, if I want to be able to search it all using Org
> mode means.  So, I segregated a handful of agenda files, and developed a
> small tool for when I need a massive org search (which is often).  Let
> me share it, in case useful to others.  For the code, see:
>
> http://stackoverflow.com/questions/6554110/emacs-outline-mode-auto-expand-upon-grep/8439425
>
> The idea is simple: just use a recursive grep and normal compile-type
> buffer and commands, yet add a bit of code to reveal hidden text which
> is being located by the compile commands.  Also manage so that a mere
> C-s on the located buffer immediately show all occurrences.

I guess your code runs faster, but you might be interested in 

  M-x org-occur-in-agenda-files RET

HTH,

-- 
 Bastien

Re: Massive Org search

[Bernt Hansen]

pinard@iro.umontreal.ca (François Pinard) writes:

> Hi, Org people.
>
> I'm a rather recent Org mode user.  Consolidating my previous data into
> Org mode gave 385 files using 5,5 Mb.  I quickly found out that this is
> too much for Org mode, if I want to be able to search it all using Org
> mode means.

Maybe it's the number of files... I have 91 files (including .org and
.org_archive) totalling 12M without any issues.

Regards,
Bernt

Re: Massive Org search

[François Pinard]

Bastien <bzg@altern.org> writes:

> I guess your code runs faster, but you might be interested in 
>   M-x org-occur-in-agenda-files RET

Unless I remember wrongly, I tried it.  With many agenda files, Org
became too slow for my needs.  So, I thought I was unreasonable, and
managed to keep only a few agenda files, and made the little code I
shown you, as a (reasonable) compromise.

I guess that with fewer files, I would not have felt any problem.

François

P.S. This morning, I created a table with nearly 2000 entries.  The
initial creation took a few seconds, but then, the updating afterwards
is much faster than I expected, it is even quite workable.  I suspect
(without knowing) that much attention has been paid to these aspects.

Re: Massive Org search

[François Pinard]

Bernt Hansen <bernt@norang.ca> writes:

>> I'm a rather recent Org mode user.  Consolidating my previous data into
>> Org mode gave 385 files using 5,5 Mb.  I quickly found out that this is
>> too much for Org mode, if I want to be able to search it all using Org
>> mode means.

> Maybe it's the number of files... I have 91 files (including .org and
> .org_archive) totalling 12M without any issues.

Hi, Bernt.  I did not have any issue either, besides the slowness.  If
you can handle 12M of agenda files briskly, maybe I should retry once
I'll have some more Org experience.  Thanks for your feedback.

François

Org-mode performance (was: Massive Org search)

[Karl Voit]

* Bernt Hansen <bernt@norang.ca> wrote:
>
> Maybe it's the number of files... I have 91 files (including .org and
> .org_archive) totalling 12M without any issues.

In the Memacs-project[1] we do face certain performance issues with
large data sets in Org-mode. Therefore we are currently developing
and running benchmarks.

I'll post here the results.

What usually helps is: keeping the vast amount of (old?) data in
.org_archive files. This speeds up daily work while giving you the
opportunity to activate the rest of it on demand.

  1. https://github.com/novoid/Memacs
-- 
Karl Voit

Re: Org-mode performance

[Karl Voit]

* Karl Voit <devnull@Karl-Voit.at> wrote:
> * Bernt Hansen <bernt@norang.ca> wrote:
>>
>> Maybe it's the number of files... I have 91 files (including .org and
>> .org_archive) totalling 12M without any issues.
>
> In the Memacs-project[1] we do face certain performance issues with
> large data sets in Org-mode. Therefore we are currently developing
> and running benchmarks.
>
> I'll post here the results.
>
>   1. https://github.com/novoid/Memacs

We've got some simple benchmark results for larger data sets within
Org-mode files:

https://github.com/novoid/Memacs/blob/master/performance.org
(Sorry, «many Org-mode files» was no issue at our side, so we did
not research into that.)

In short:
* Best performance: only without any drawers at all
* Acceptable performance when using another drawer than :PROPERTIES:
* Worst performance when using :PROPERTIES: drawer

So if you want to have many entries in Org-mode files, avoid drawers
if you can. If you need a drawer, probably choose to define your own
drawer.

-- 
Karl Voit