Re: [Tracker] Large sparql query results

[Philip Van Hoof <spam pvanhoof be>]

On Thu, 2010-03-04 at 22:07 -0500, Spivak, Max wrote:

Hi Max,

> I have a question about large query results from tracker-store. in my
> experience with SQL databases and their client api, when a query
> results are generated, only a subset are initially transferred to the
> client. The rest are loaded on demand as the client iterates over the
> results. 

This usually only is the case when the client requests iterating over
the result using a so called cursor.

Usually are those cursors also one-direction only (you can't go back).

Because SQLite isn't MVCC it would mean that for the whole duration that
a client keeps such a cursor open, all other database connections will
be more or less blocked from doing things.

This is why we don't support a cursor API.

You can emulate a cursor API with LIMIT and OFFSET, of course.

> How does libtracker-client handle this? What if the result set is too
> big to fit in the client memory? Is there any sort of incremental
> result returning ability or callback support with partial results? t

We at this moment don't provide this ourselves in libtracker-client.

LibQtTracker has a "streaming model" which will do this for you on a
QAbstractDataModel as far as I know.

http://maemo.gitorious.org/maemo-af/libqttracker


Cheers,

Philip

-- 
Philip Van Hoof, freelance software developer
home: me at pvanhoof dot be 
gnome: pvanhoof at gnome dot org 
http://pvanhoof.be/blog
http://codeminded.be