A new semantic search engine for the KDE desktop
Overwhelming
Jos Poortvliet also admits, however, that the new search "overwhelms" the I/O on certain systems. Although the new indexer can index multiple gigabytes in a few minutes, it chokes on very large files. As an example, Jos Poortvliet cited text files of more than 20MB. The indexer classifies such files as "bad files" and then no longer touches them. But, to forge its way faster through the file set, the indexer always processes 40 files at one go.
Baloo from KDE Applications 4.13 uses a timeout of five minutes. Once this is exceeded, the indexer divides the 40 files into two groups. It then again tries to index each of the two groups. After working for five minutes on one of the two groups, it also divides this group. The whole process continues until the indexer has found the rogue file.
However, the long timeout keeps the indexer busy for a relatively long time. The developers have therefore reduced the timeout to two minutes, and the indexer now handles problematic files in a better way. However, the changes are unlikely to reach most users until the next version of KDE Applications.
Configuration from a User
Lindsay Mathieson has improved the rather terse configuration dialog [13]. When this issue went to press, however, it was not yet known whether, when, or how it would end up in KDE Applications. In contrast to the current version, it not only lets you switch off the indexer, but also provides an easy way to define exception rules for directories, files, and MIME types (Figure 5). The KDE developers are looking to expand the Baloo search service and add more applications. For example, a balooctl
command-line tool will start and stop the service (similar to nepomukctl
and akonadictl
).
Roadmap
The plans also include a spellchecker, support for removable disks and expanding the Milou plasmoid to a complete search [14]. In addition, the KDE developers are working on porting Baloo to KDE Frameworks 5 This version is designed to be fully compatible with the current version.
The KDE developers are pushing forward with Baloo. Users and programmers will have to make friends with semantic search. Only time will tell – when more KDE applications start to use it – whether it eventually becomes genuinely fast and stable.
Infos
- Nepomuk project: http://nepomuk.semanticdesktop.org
- RDF: http://www.w3.org/RDF/
- Vishesh Handa, "Nepomuk in 4.13 and Beyond": http://mail.kde.org/pipermail/nepomuk/2013-December/004858.html
- Jos Poortvliet, "KDE's Next Generation Semantic Search": http://dot.kde.org/2014/02/24/kdes-next-generation-semantic-search
- Baloo: https://community.kde.org/Baloo
- Baloo's architecture: http://community.kde.org/Baloo/Architecture
- (KDE) applications ported from Nepomuk to Baloo: http://community.kde.org/Baloo/NepomukPort
- Ask Ubuntu – "How to Turn Off Baloo in KDE 4.13?": http://askubuntu.com/questions/437635/how-to-turn-off-baloo-in-kde-4-13
- KDE-Forum – "How Do I Pause Indexing?": https://forum.kde.org/viewtopic.php?f=154&t=120047#p304335
- Baloo discussions on the SUSE mailing list: http://lists.opensuse.org/archive/opensuse-kde/2014-05/msg00008.html
- "Teacups and Storms!" by Jos Poortvliet: http://blog.jospoortvliet.com/2014/05/teacups-and-storms.html
- Call for testing: http://dot.kde.org/2014/03/12/applications-413-coming-soon-help-us-test
- Revised Baloo dialog: https://gitorious.org/baloo-kcmadv
- Tasks: http://community.kde.org/Baloo/Tasks
« Previous 1 2 3
Buy this article as PDF
(incl. VAT)