Re: New module proposal: tracker

[Sandy Armstrong <sanfordarmstrong gmail com>]

On Thu, Oct 29, 2009 at 3:59 PM, Martyn Russell <martyn lanedo com> wrote:
> On 29/10/09 22:49, Sandy Armstrong wrote:
>>
>> On Thu, Oct 29, 2009 at 3:06 PM, Patryk Zawadzki<patrys pld-linux org>
>>  wrote:
>>>
>>> On Thu, Oct 29, 2009 at 10:51 PM, David Zeuthen<david fubar dk>  wrote:
>>>>
>>>> Not to sound like an asshole or anything but, I mean, didn't distros try
>>>> including Tracker by default in previous releases? AFAIR, it didn't
>>>> really work well. So if GNOME included Tracker in the next release and
>>>> core parts of GNOME started depending on it in a way that couldn't be
>>>> turned off... then distros would be in a lot of trouble if Tracker
>>>> didn't work well. This would probably end up reflecting badly on the
>>>> GNOME project.
>>>
>>> Are we talking about the tracker crawler or tracker itself (index and
>>> metadata storage)?
>>>
>>> I can understand how including the crawler is not the best idea
>>> (especially until we get ourselves recursive inotify support) but what
>>> can go wrong with tracker itself?
>>
>> What can go wrong introducing a brand new barely-used technology and
>> set of APIs as something we call "GNOME"?  I think that question
>> answers itself. ;-)
>
> I already said which applications use it. Tracker is not brand new and it is
> not barely used.

To be clear, I was speaking about the store as opposed to the
indexing.  Where is it used on the desktop?

The 0.7.x APIs are brand new and barely used, if I've understood this
thread correctly.

>> So the most pragmatic approach seems to be to wait until
>> Tracker matures and people are using it in ways that make including it
>> as part of the GNOME desktop an easy choice.  Right now, we're still
>> at the "imagine the possibilities" stage with Tracker on the desktop.
>
> Are we? Have you tried it yet?

Looking a few dozen emails back at your list of how Tracker is
currently used, everything currently implemented is related to
indexing, not the store (unless you look at Maemo).

Also, your list was a list of applications that somehow use or
integrate with Tracker.  It was not a list of use cases, and did not
describe in detail how those applications are integrating with
Tracker.  Is it just more indexing stuff?

And no, I have not tried Tracker in a long time.  An indexer is not
something useful to me (though I agree it is useful to many people),
and the store has (at this time) nothing to offer me as a user, from
what I understand.

>> I'd rather wait until some compelling use cases are actually implemented.
>
> What use cases would you like to see?

I think this is kind of the core of the problem here.  I get the
feeling that the Tracker developers are proposing a technology because
they want to enable application developers, and that it is expected
that application developers will come up with and implement compelling
use cases.  Like others in this thread, I think this is just
backwards.  I feel like the applications and use cases should shape
the Tracker API, and it's in everyone's best interest not to bring it
in prematurely.

I'm not trying to knock Tracker at all, because I think it's cool
stuff.  I just would rather we bring it in when our applications are
using it in ways that will provide clear benefits to our users.

Right now, the only clear benefit I see is better search thanks to the indexer.

Sandy