Re: [Tracker] [Fwd: Plans for 0.7]




On Wed, 2008-03-05 at 10:03 +0000, Martyn Russell wrote:
Jamie McCracken wrote:
On Tue, 2008-03-04 at 09:40 +0100, Nikolai Weibull wrote:
On Mon, Mar 3, 2008 at 5:24 PM, Jamie McCracken
<jamie mccrack googlemail com> wrote:

 From: Ivan Frade <ivan frade nokia com>
  I prepared a small diagram to clarify the intended architecture for
 tracker 0.7. I think that a diagram like this can be very useful to
 understand all required refactorings.
Does the addition of indexing of removable media affect this in any
way?  I would really like to see that added.  I'd gladly help out, as
I've already mentioned, but the only real response I got back then was
that there were a couple of ideas for how that was to be added, but
nothing concrete.

we will definitely expand hal support for removable media

I have actually got a HAL patch which does this. There is one problem
with the implementation of our HAL patch and the tracker design - it
stores things based on file location. This is all well and good for a
lot of things, but when it comes to removable media really you can't
just count on the path. You really need to use the device ID which is
HAL can tell us about.

What is needed as far as I understand it, is to improve the database
storage to include such information so we know if device foo mounted on
/media/sdcard is actually the same as device bar which may also be
mounted in the same place.

This would greatly improve searching removable devices (after their
initial indexing of course).



db already has these facilities

we have volume table for UDI -> URI
http://svn.gnome.org/viewvc/tracker/trunk/data/sqlite-tracker.sql?view=markup


We have service stable for rest-:
services.AuxilaryID is to be used for link to Volumes.ID
Services.Enabled is used to indicate if hit is available (IE mounted)
http://svn.gnome.org/viewvc/tracker/trunk/data/sqlite-service.sql?view=markup


just need to tie it all together. 


Indeixng needs to be changed to set above and to enable/disable entries whenever relevant volume is 
mounted/unmounted.

All search sql needs to be adjusted to ignore hits when Enabled = False

jamie





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]