Re: Beagle Inclusion/Exclusion Lists

[Joe Shaw <joeshaw novell com>]

Hi,

On Sat, 2006-04-08 at 19:34 -0500, Richard Laager wrote:
> 1. Does Beagle have a default exclude list of things like *~, *.o, etc.?

See Daniel's email about this.  Note that this *only* affects the file
system backend.  Gaim has its own backend, so if you're thinking in the
context of it, you'd need to handle this yourself.

> 2. How well does Beagle cope with removable media? For example, if I
> add /media/usbdisk as a folder to search, is it going to drop and
> reindex that every time the disk is unmounted and remounted? Would I be
> better off using static indexes? I'd really like it to auto-update,
> though. Maybe I could use static indexes and there's some way (via
> gnome-volume-manager or whatever it's called) to re-index when the disk
> is connected?

By default, we only search your home directory.  Most distros that ship
beagle also have a cron job which uses static indexes for applications,
system documentation, and sometimes windows partitions.  So in most
cases, removable media isn't ever indexed.

Beagle today doesn't do any special handling of removable media.  So the
end result is difficult to predict.  But what I believe would happen is
this: 

        * You'd add the removable media as a separate root.  Assuming
        it's attached, it would be indexed as part of the standard
        initial crawl.
        
        * When you unplugged it, basically nothing would happen because
        Beagle doesn't deal with the "unplugged" inotify event.
        
        * If you did searches, items wouldn't show up even though they
        were still in the index because the file system backend does one
        last "does the file exist" check before returning results.
        
        * If you plugged the media back in and redid the search, you
        would see the results.
        
        * It's not clear to me what would happen if you shut down the
        daemon and restarted it with the media removed.  It's likely the
        contents of that directory would be removed from the index
        because they're no longer there on disk.

You could use static indexes for this.  There is a utility written by a
contributor which actually uses inotify events to rebuild static
indexes, which is in bugzilla.  You could use that, but it's definitely
a hack for your use case.

It's something which we need to address.

Joe