update-db cron job: solving a long-standing issue

Milan nalimilan at club.fr
Mon Sep 17 18:47:04 UTC 2007

Mark Schouten said:
> I prefer this too. I also think it is good to think about newbies, but
> is it really necessary to ignore more advanced users just because they
> know what they're looking for? I know I would be annoyed if locate was
> missing on my server.
We're not talking about servers but only Desktop versions. Of course, on
servers admin should need it.

Note I'm not hating locate by principle, but because it makes sometime
computers hang without explanation. If we could use a more comprehensive
way of indexing files, like Tracker does (ie when you do'nt work), this
could be OK. Comparison with Tracker is not accurate because of this
rlocate seems to be resource-intensive too, because it needs a complete
rescanning every 10 starts or so. IMHO, a workaround with find and dpkg
is not so bad for occasional usages, and 'apt-get install slocate' is
easy for anybody using the command-line.

Colin Watson said:
> Can we not come up with a way to generate the locate database
> from tracker instead?
Beagle does this for system-wide documentation, AFAIK. So this is
possible, only taking care of the filenames. (But Beagle was eating CPU
doing this too, though it is not necessary.)

The dependencies point should be investigated more, but AFAIK
gnome-utils (ie gnome-search-tool) doesn't depend on locate. Is it able
to use find ?

Anyway, I've opened a bug here:
We should use it when we have found a common position.

More information about the Ubuntu-devel-discuss mailing list