* Tim Gray on Sunday, May 08, 2011 at 17:10:22 -0400
> On May 08, 2011 at 10:47 PM +0200, Christian Ebert wrote:
>> $ time mairix -v -p
> 
> I bet that was my problem.  I don't think I ever used -p, so there
> were a lot of dead messages floating around in my db.

I believe --purge makes it actually slower, as it compacts dead
messages away in the db.

~$ time mairix

real    1m54.450s
user    0m28.735s
sys     0m54.465s
~$ time mairix --fast-index

real    0m37.570s
user    0m28.725s
sys     0m1.503s
~$ time mairix --fast-index --no-integrity-checks

real    0m28.202s
user    0m26.776s
sys     0m0.772s
>
> The times I'm getting now are pretty good.  Notmuch seems to be
> faster, but the times are all low enough that I don't have a problem
> with any of them.
> 
> mairix -v -p
> ------------
> real  0m17.682s
> user  0m4.911s
> sys           0m8.524s
> 
> notmuch new
> -----------
> real  0m5.152s
> user  0m0.067s
> sys           0m0.261s
> 
> Searches for the two showed a similar gap.  Again, neither was slow
> enough for me to lose any sleep over.
> 
> mairix: 0m3.044s
> notmuch: 0m0.410s
> 
> This is an interesting discussion though.  I might play around with
> mairix a bit more again.  I still see mu and notmuch having a major
> advantage of being built on proper database tools.  I get a lot of
> errors about messages not being indexed by mairix, and that whole
> recommended dance of removing the lock file before a search, etc. is
> annoying as well.  Furthermore, the thing that excites me about
> notmuch that the others don't have is the fact that it's built as a
> library.  An enterprising developer could integrate it into a mail
> client (other than the emacs thing they have going on) and it would be
> pretty great in my mind.  Remember, notmuch isn't just an indexing
> tool - it also lets you tag messages and search on tags, etc.

I would rather be interested whether notmuch correctly finds
strings containing non-ascii characters. In my script I work
around by appending the results of a second search, so when I
enter e.g. 'première' in utf-8, the first search is in utf-8 and
second, usually finding more matches converts the search string
to cp1252 - but that is obviously a hack.

I once tried with mu, and it wasn't any better in detecting
those, but perhaps I gave up to quickly.

c
-- 
\black\trash movie   _SAME  TIME  SAME  PLACE_
               New York, in the summer of 2001

--->> http://www.blacktrash.org/underdogma/stsp.php

Reply via email to