After more than a year of heavy work in tracker, here is the brand new tracker release, 0.6.90!!
If you are still wondering what tracker is: it is a daemon monitoring the filesystem, extracting metadata from your files and indexing them (and the contents of text files) so that you can search fast and efficiently for whatever you need. The point of the metadata is that you can query pretty interesting things like “all songs from the artist X” or “all images from last week tagged as ‘holidays'” without worrying about the physical location/name of the file (as long as your ID3/Exif is correctly set ;))
In this last year, we refactor (well, almost rewrote) the daemon, split it into two processes, enabled the dbus-introspection, added few methods in the API, replaced the homemade crawler with GIO and cleaned the code in general. Now it should be faster and more responsive, and it should also be much easier to extend it to support new categories and files. You can even write indexer modules without compiling the whole tracker! If you want more details, the Changelog is full of them.
For the maemo fans, the good news is that 1) this version is pretty much what you are going to find in fremantle 2) It is the same code in maemo and in desktop, so… you can test it on your desktop and if you find/report bugs it will help desktop and maemo users: two happy platforms for the same price! You can fill your complains and suggestions in the upstream gnome.bugzilla, in the mailing list, or chatting directly with us in the IRC (#tracker in gimpnet).
Our intention is to start releasing more frequently… though it looks like “i’ll be posting in the blog more frequently” 🙂 There are a lot of interesting things coming in the near future: RDF, sparql, nepomuk, more applications pushing data, more complex queries,… and it is going to be so interesting having this running on a mobile device!

How much of a performance and battery life hit will it be to run Tracker as a background process on a tablet?
I don’t have serious measures about this topics, though as a user, I didn’t perceive any difference in battery life. We have also fine-tunned tracker to keep the system always responsive.
The background process running all the time just keeps the filesystem watches and the DB Connections to answer queries.
All the other heavy work (indexing and extracting metadata) is done in different processes, which can be stopped/paused as needed (because the device is running out of battery, for example).