2010-06-10 Martyn Russell Updated NEWS 2010-06-10 Jürg Billeter tracker-miner-fs: Do not delete metadata from other sources on update When a file is updated, only delete statements that have been inserted by the miner instead of relying on the inflexible tracker:isAnnotation. Fixes NB#173164. tracker-miner-fs: Use DELETE instead of DROP GRAPH for volumes Our DROP GRAPH implementation was not meant to be used like that. Regular DELETE semantics work fine here. libtracker-data: Never set graph for internal property tracker:modified libtracker-data: Add test for GRAPH in predicate variable queries SPARQL: Only retrieve graph from database when necessary SPARQL: Support GRAPH in predicate variable queries libtracker-miner: Revert deletion of original resource on update This reverts commit cb0c4c59e60294a455be8f4263dc1fafdc27527c. The real bug was in IRI cache invalidation, which is now fixed. File update should not trigger a complete resource deletion, it should only delete embedded metadata. 2010-06-10 Mikael Ottela Updated generator scripts to include dataobjects for emails. 2010-06-09 Martyn Russell libtracker-db: Moved last-crawl.txt machinery to here from miner-fs * This was previously called crawling-timestamp.txt. It has been renamed last-crawl.txt now. * Fixes GB#620353, remove crawling interval timestamp on re-index * Now libtracker-db removes this file too when a reindex is requested libtracker-db: Updated first index filename and set contents Use first-index.txt not .firstindex Set content to version that did first index Fixed broken whitespace/tabs in configure.ac 2010-06-09 Patrick Ohly --enable-maemo: had no effect enable_maemo was never set to anything but "no" in the default case in configure.ac. Note that enable_maemo=yes in the "enable option selected" case would be wrong, because --disable-maemo is valid and ends up in that branch with $enableval == "no". Some other uses of AC_ENABLE() make that mistake (not fixed in this patch). 2010-06-09 Carlos Garnacho TrackerMinerFS: Fix IRI cache invalidation. When a NULL value was being inserted in the hashtable, the next query for it is supposed to issue a item_query_exists() call, but the iri was only being stored in the hashtable, and not returned so its used. 2010-06-09 Philip Van Hoof tracker-extract: Fixes NB#172414, check for empty string 2010-06-08 Philip Van Hoof libtracker-miner: Fixes NB#161449, on overwrite the original resource must be deleted Fixes NB#173206, Also merge manufacturer correct for videos 2010-06-08 Jürg Billeter libtracker-data: Add test for FILTER in group graph pattern SPARQL: Always generate SELECT around group graph pattern Without extra SELECT, invalid SQL may be generated for certain FILTER clauses. 2010-06-08 Philip Van Hoof Fixes NB#173206, use nfo:device and merge make and model correct for videos 2010-06-07 Philip Van Hoof tests: Add to unit test case tests: Add unit test for Bugfix NB#172652 libtracker-data: Bugfix NB#172652, comparing NULL isn't reliable 2010-06-07 Andrej Žnidaršič Updated Slovenian translation 2010-06-07 Jürg Billeter SPARQL: Warn instead of crash on namespace without prefix icons: Fix make distcheck tracker-miner-flickr: Fix make distcheck SPARQL: Support experimental INSERT SILENT and DELETE SILENT When the SILENT keyword is used, non-syntax errors will be ignored. libtracker-data: Clear buffer only when rolling back transaction libtracker-data: Fix error annotations in Vala bindings maemo: Set default value for maemo:relevance to 1000000 libtracker-data: Add default value support 2010-06-07 Aleksander Morgado libtracker-common: remove public gtk-doc documentation 2010-06-05 Gabor Kelemen Updated Hungarian translation Updated Hungarian translation 2010-06-04 Ivan Frade Functional tests: BatchCommit + kill the daemon Fixes NB#171657, Device manufacturer and model XMP data not extracted mp4 2010-06-03 Jorge González Updated Spanish translation