2010-02-19 Martyn Russell Updates NEWS with last minute changes libtracker-db: Bump DB version for localtime/named graph changes 2010-02-19 Philip Van Hoof tracker-store: Emit class update signal for each rdf:type 2010-02-19 Jürg Billeter libtracker-common: Fix C89 warnings libtracker-data: Test local time support SPARQL: Support fn:*-from-dateTime functions to retrieve local time libtracker-data: Insert and update localDate and localTime columns libtracker-common: Add TrackerDateTime to store time with UTC offset libtracker-data: Add localDate and localTime columns 2010-02-19 Martyn Russell Merge branch 'hierarchical-indexing' libtracker-miner: Fixed finishing message alignments 2010-02-19 Carlos Garnacho TrackerMinerFiles: do not monitor child dirs of IndexSingleDirectories. Direct child dirs were being added a monitor, even if the crawler didn't get in there. TrackerMinerFS: Fix progress reporting It also needed to take into account the items left in the directory trees pending insertion into the items_created list. TrackerCrawler: honor the recurse parameter. 2010-02-19 Martyn Russell Updated NEWS with tracker-tag last minute fix Updated NEWS with translations Updated NEWS 2010-02-18 Mario Blättermann Updated German translation 2010-02-18 Petr Kovar Update Czech translation by Marek Cernocky & Petr Kovar Reviewed by: Lucas Lommer. 2010-02-18 Martyn Russell libtracker-miner: SPARQL performance improvement Also use " not ' to be consistent libtracker-miner: Make queue enum a typedef for type checking libtracker-miner: Free nodes using G_POST_ORDER So we free children before parents libtracker-miner: Use useful #ifdef for tree debugging not #if 0 libtracker-miner: Only set parent after no errors are returned libtracker-miner: Avoid unnecessary cast libtracker-miner: Don't use g_slice_new0 for all EnumeratorData members libtracker-miner: Use a GSList in the crawler for DirectoryProcessingData 2010-02-18 Carlos Garnacho Add missing user_data parameters to TrackerCrawler::finished callbacks. Use correct signature for TrackerCrawler::finished. TrackerMinerFS: Plug small leak. TrackerMiner: Cancel pending async calls on dispose. GSimpleAsyncResult tries to reference the object (the miner in this case), so it likely crashed when trying to ref an object being destroyed. Instead do it on dispose, when the object is still alive, but about to be destroyed. 2010-02-18 Dariusz Wiechecki (EXT-Comarch) tracker-extract: Use artist's value in filename when creating new album art 2010-02-18 Jorge González Updated Spanish translation 2010-02-18 Jamie McCracken TST- made metadata tile labels category specific added folder display 2010-02-18 Carlos Garnacho TrackerMiner: Cancel pending async calls on dispose. GSimpleAsyncResult tries to reference the object (the miner in this case), so it likely crashed when trying to ref an object being destroyed. Instead do it on dispose, when the object is still alive, but about to be destroyed. TrackerMinerFS: if0'ed out debug code. TrackerMinerFS: Avoid safely ignored root directories. This may happen on reindex. 2010-02-18 Dariusz Wiechecki (EXT-Comarch) tracker-extract: Use artist's value in filename when creating new album art 2010-02-18 Carlos Garnacho TrackerMinerFS: Ensure ::check-directory is called on root folders. TrackerMinerFiles: Use TrackerMinerFS API to get URNs. Now both file and parent URNs are taken from TrackerMinerFS if any. In the case a file doesn't exist (i.e. doesn't have an URN yet) an anonymous URN will be used as usual. TrackerMinerFS: Add API to know a file/parent urn. This API allows retrieving the URN for a GFile being processed (to check for file existence in the store), or it's parent's (to set nfo:belongsToContainer relationship). Since files are guaranteed to be processed together with the other files in the same folder, and parent folders are guaranteed to be fully processed before its children, The method to cache the parent folder URN should have little impact on performance. In order to retrieve the URN for the file itself, a extra call to item_query_exists() had to be added, unfortunately, doing so on should_change_index_for_file() is not an option, since it happens on the crawling phase, so all URNs would have to be stored. This is a performance vs memory usage problem. TrackerMinerFS: warn if one file has >1 URNs. TrackerCrawler: Report a GNode with the file hierarchy. TrackerCrawler now provides a GNode with the directory/file hierarchy on ::directory-crawled for each GFile passed at tracker_crawler_start(), all the stats are also reported per directory root. ::finished is now only used to notify about interruption. TrackerMinerFS now takes a copy of each of the GNodes intead of appending directly to items_created. Whenever items_created is empty, the stored GNodes are checked for more contents to inspect. This is done so TrackerMinerFS guarantees that a directory has been already added to the store before processing its children. Tracker-extract: Do not specify subject in embedded insert. TrackerSparqlBuilder: allow embedder to specify the subject. Embedded inserts now only consist of predicate/subject, so the embedder is able to specify the subject the data is applying to. 2010-02-18 Jürg Billeter libtracker-data: Fix DROP GRAPH after switch to use nie:url 2010-02-18 Mikael Ottela Add some requested properties to be full-text-indexed. Add fn:concat Add the implementation for the xpath function concat and unit-tests. Add fn:string-join Add the implementation for the xpath function string-join and unit-tests. Support NULL values in functions. 2010-02-18 Martyn Russell Updated NEWS Fixes NB#157277, Not able to search for files based on a particular tag Don't use regex matching for searching by tag name, use exact matching Bumped version to 0.7.21 Updated NEWS tracker-tag: Removed unused code and fixed memory leak when removing tags Fixes GB#610318, Bump GTK+ dep to 2.18 due to gtk_widget_set_can_default() Tests: Restructured data and added metadata extraction test 2010-02-17 Martyn Russell nautilus: Don't show an empty file count when adding new tags Fixes GB#610026, Nautilus Tagging Extension does not work at all. 2010-02-17 Jürg Billeter libtracker-data: Update tests for named graphs with new syntax SPARQL: Support INSERT { GRAPH <...> { ... } } Syntax specified in SPARQL 1.1 Update draft. libtracker-db: Update journal to support named graphs libtracker-data: Add test for named graphs with properties libtracker-data: Add insert-delete-insert test for named graphs libtracker-data: Test simple named graph insert and query SPARQL: Support simple named graph patterns in queries libtracker-data: Insert and update graph columns libtracker-data: Add graph column next to each property column 2010-02-17 Martyn Russell tracker-extract: Warn if a module doesn't export the required functions 2010-02-17 Michele Tameni Fixes libtracker-extract documentation example which was incorrect 2010-02-16 Jamie McCracken TST: Fixed display of type and formatted file size in metadata tile cleaned up meatdata tile display 2010-02-15 Martyn Russell tracker-tag: Fix looking up by filename Use g_file_new_for_commandline_arg() and also fixed SPARQL. tracker-search-tool: Fix failure to build with Vala 0.7.6. Was due to commit 6f991e3e3d25d043a9e4048f3b0085d1de467ba9 2010-02-15 Jürg Billeter tracker-extract: Link all extractors to libtracker-extract Util functions have been moved from libtracker-common to libtracker-extract. tracker-info: Print queried URN 2010-02-15 Jamie McCracken TST: Added gradient effect to Metadata Tile 2010-02-12 Mario Blättermann Updated German translation 2010-02-12 Philip Van Hoof tracker-client: Refactored away tracker-statement-list tracker-extract: Moved util functions from libtracker-common to libtracker-extract Fixed compilation warnings, mixing code with variables 2010-02-12 Jürg Billeter libtracker-common: Fix header filename for TurtleWriter 2010-02-11 Michael Biebl Link kmail plugin against libtracker-client for tracker-sparql-builder