Re: [Tracker] How do all the components fit together

[Ivan Frade <ivan frade gmail com>]

Hi!

 (Readding the mailing list in CC)

On 8/19/11, James Hurford <terrasea gmail com> wrote:
> Am I correct in this?
>
> - miners are independent of the rest of tracker

 Yes, they are independent processes and you could write them without
any dependency to Tracker (any process pushing sparql into tracker can
be considered a miner).

 That being said, to make the things easier miners should implement
the interface in libtracker-miner, and use libtracker-sparql to insert
stuff into Tracker. So we strongly recommend to use those Tracker
libraries but are not mandatory to have a miner.

> - miners don't know about each other and neither does tracker know about
> them

 Yes (I assume "tracker" means "tracker-store")

> - miners only purpose is to extract data from various sources and lets
> store know about this data

 Yes.

> - the store is responsible for storing this in some sort of DB

 Yes.

> - tracker-extract is only used by tracker-miner-fs to extract meta
> data about the files it finds

 Yes

> - is tracker-extract used by any of the other miners?
 No, it is not used by other miners

> - if not is it possible for other miners to do so?

 Nothing prevent them to do so. You could write a new miner for some
specific purpose and call tracker-extract when needed. I wonder if
this situation could ever happen... usually fine-tuning the miner-fs
configuration and adding few extractor-module solve anything related
with local files.

 Regards,

Ivan