Re: Proposing Tracker for inclusion into GNOME 2.18

[Jamie McCracken <jamiemcc blueyonder co uk>]

Jeff Waugh wrote:
> <quote who="Jamie McCracken">
>
> > > Jamie, can you please explain what exactly Tracker is and does?  I have
> > > no idea what ‘first class object database’ means, so please explain what
> > > you mean by that if you need to.
> > http://live.gnome.org/ResearchAndDevelopment/Tracker
>
> Jamie,
>
> You're proposing the addition of an entirely new technology to GNOME, with
> the intention that at some stage it could enter the Platform and be used by
> many (if not almost all) applications. You need to be clear, succinct, and
> explain *what* it is, what it *does* and *why* it is important.

As this stage I am simply proposing tracker-search-tool as a replacement 
for the gnome-search-tool as I believe it does a better job with faster 
instant search and search snippets.

The other stuff is mainly my future plans and I already have interest 
from Epiphany (for their cool new bookmarks/history stuff - see the 
bottom of http://live.gnome.org/Epiphany/BookmarksHistoryIntegration) 
and Rhythmbox for their common music database and will be implementing 
something for them in due course.

I am of the impression that it will only becomes part of the platform if 
the above integration takes place and respective maintainers approve all 
this but yes it would be good idea to improve the communication to make 
that easier :)


> Think elevator pitch and use cases, rather than essay and hypotheticals.
>
> During this discussion, you've pitched Tracker as an indexer competitive
> with Beagle, then as a "first class object database" possibly in line with
> the old Storage ideas, then suggested that the indexer was secondary to the
> "first class object database". This doesn't really raise confidence in your
> ability to define the problem space, let alone solve the problem.
>
> Great technology loses out to poor communication (be it directly social or
> "marketing") all the time. If you want to ensure your technology doesn't
> fall into a coma of irrelevance, you *need* to step up and communicate it
> well.

yes thats *great* advice!

I will start out by mentioning why I wrote Tracker.

The main reason was I didn't like the way GNOME uses loads of different, 
inefficient and incompatible means of storing information (think 
Berkeley DB for EDS, MBox for emails, the zillions of small performance 
draining XML files used for bookmarks, history, rhythmbox's music 
database and many other things). So, I wanted to bring together all this 
stuff under one centralised database and in doing so increase 
performance, power and memory efficiency of the platform as a whole.

There are two ways to do this - either by externally indexing this 
content, duplicating the data and acting as a high speed cache or by 
providing the actual storage for it. Tracker provides for both of these 
but the latter is preferable for efficiency reasons.

With a centralised DB, it is easy to index, make metadata extensible, 
apply tags and contextual linking to all this data and make tracker a 
high level data hub for accessing all this information from one source.

I also wanted files to be indexed so we could create flat views of data 
so we could say "show me all files of a certain mime type or a certain 
service" like all music files because I currently find the existing file 
dialogs inefficient to work with their pokey little windows and painful 
hierarchical views - we need more intelligence so when opening an office 
document it can show you all office documents on your system instead of 
only in a path

Common db storage is also one way to sort out the mess of EDS, GAIM (and 
possibly telepathy?) all using different means to identify a "person".

So tracker is basically a key part of creating a really well integrated 
desktop with lots of potential to make GNOME a more intelligent and 
productive environment for its users.

I am sure there are many future possibilities for this too but for now 
Tracker is learning to walk first before trying to run and hence my 
decision to ease it in now with the much needed search technology that 
GNOME needs to compete with Vista and OS/X.

It would be fantastic for Gnome 2.18 to have this around the time Vista 
ships and with a common dbus interface for indexing, Beagle too can 
benefit by effectively being in too for those that want more indexing 
than tracker currently provides (I dont plan on indexing as much as 
Beagle or maybe Strigi so there is bound to be room for them too).


--
Mr Jamie McCracken
http://jamiemcc.livejournal.com/