Biweekly Tracker development planning

We realize

At the Tracker team we realize that we should not only tell you guys about Tracker’s new features. We should also involve you in the planning and realization of software development at the Tracker project.

I have discussed this with our team and we agreed that I will make a summary at each beginning of our biweekly planning.

The reason is fairly simple: we want to involve the community and inspire them to become active contributors. But how can they contribute unless they know what the plans of the development team of Tracker for the next few days are? You are right, they can’t.

Which is a situation that we want to change. Therefor, here’s the summary of the plans for the next two weeks:


  • Making tickets for an audio ontology that we made during a previous sprint to upstream Nepomuk.
  • Craft an ontology for images that will later be proposed to Nepomuk. We will convert the extractors in our experimental branch to use the new image ontology.
  • Applications need to be able to create arbitrary key value pairs to existing resources. We plan to add a custom ontology on top of Nepomuk’s NAO ontology that will make this possible.

Fixing things in the 0.6.9x series

  • Changing unique value sorting to use collations and we’re planning to change the unique value queries for better performance.
  • Making sure tracker-indexer and tracker-extract support detecting unmount events, and marking resources as unavailable, at a very early stage. This helps cleanly unmounting a removable device without keeping the device busy as the unmount request takes place.
  • Supporting listening for a signal that tells us to pause the indexer. This will be helpful on devices that want us to be silent while they are performing a higher priority task.
  • Evaluating and testing concurrent access.

Our experimental branch

  • Start discussing merging our experimental branch to trunk together with Jamie.
  • We have updated the new .ontology files in our experimental branch to include more specific rdfs:range information.
  • Making it possible to observe changes about resources that belong to a class. For example observing changes in all nfo:Document classes. The granularity will be at the level of a resource. You’ll know whether a resource is updated, deleted or created. This task might take more than a week to complete. Ivan Frade is planning to write a blog item dedicated to this subject.
  • To fulfill the requirements of certain use-cases we’re planning to make it possible to delete a complete resource (or a set of) with one function call.
  • We might start working on adding support for application domain (custom) ontologies.
  • Porting recent improvements on support for detecting file renames to our experimental branch.
  • Evaluating isLogicalPartOf or alternative link as cascading rule for deletes.

When people are interested in joining development of Tracker, they can ping us at the channel #tracker on GimpNet. They can also send a E-mail to Tracker’s mailing list.