| 495402f | Dan LaRocque | 10 October 2014, 02:48:18 UTC | [maven-release-plugin] prepare release 0.5.1 | 10 October 2014, 02:48:18 UTC |
| ef725f7 | Dan LaRocque | 10 October 2014, 02:45:42 UTC | Updating titan.compatible.versions prop | 10 October 2014, 02:45:42 UTC |
| 2411f03 | Dan LaRocque | 10 October 2014, 02:39:42 UTC | Revert CommonsConfiguration to Enum.toString usage This commit effectively reverts 4c2c34226686ed91feb230af5c8a7a8e39b70403, but it also changes Timestamps.toString just delegate to the name method, so the strings stored in CommonsConfigurations backers will still be e.g. MILLI rather than Timestamps[ms]. Like 4c2c34226686ed91feb230af5c8a7a8e39b70403, this is orthogonal to KCVSConfiguration which uses a StandardSerializer to convert between enums instances and StaticBuffers. | 10 October 2014, 02:41:04 UTC |
| 579301b | Dan LaRocque | 10 October 2014, 00:57:15 UTC | Update upgrade instructions for 0.5.1 Copied @mbroecheler's UPGRADE.asc additions on 21478ebeea4f5df27b757373b3bf5dc046c34e29 to docs/upgrade.txt, added a date | 10 October 2014, 00:57:15 UTC |
| 1bc4136 | Dan LaRocque | 10 October 2014, 00:55:21 UTC | Update upgrade instructions for 0.5.1 | 10 October 2014, 00:55:21 UTC |
| 3426482 | Dan LaRocque | 10 October 2014, 00:53:53 UTC | Update changelog for 0.5.1 | 10 October 2014, 00:53:53 UTC |
| 66d86db | Dan LaRocque | 10 October 2014, 00:51:21 UTC | Update compat test to set index.search.directory This probably should not be GLOBAL_OFFLINE in the first place | 10 October 2014, 00:51:21 UTC |
| a6e751c | Dan LaRocque | 09 October 2014, 23:10:18 UTC | Merge branch 'titan05' of github.com:thinkaurelius/titan into titan05 | 09 October 2014, 23:10:18 UTC |
| 7e48baf | Dan LaRocque | 09 October 2014, 21:49:47 UTC | Reflections tweaks for 0.9.9 compatibility For #778 | 09 October 2014, 21:49:47 UTC |
| 4026360 | Dan LaRocque | 09 October 2014, 21:48:31 UTC | Drop Solr from Hadoop 1 distribution Solr is still included in the Hadoop 2 distro. Solr depends on some Hadoop 2 artifacts. This commit also defines a pom property in titan-dist-hadoop-1 that's necessary to bring in the stock Titan config files. | 09 October 2014, 21:49:00 UTC |
| 4c2c342 | Dan LaRocque | 09 October 2014, 10:23:39 UTC | Fix enum name handling in CommonsConfiguration CommonsConfiguration assumed that toString returned the same thing as name, but this is generally untrue (an untrue in particular for the Timestamps enum). | 09 October 2014, 10:23:39 UTC |
| 08bc081 | Dan LaRocque | 09 October 2014, 10:23:23 UTC | Cleanup/expand some ConfigOption descriptions Tidying up some setting descriptions that are now included as comments in the stock configuration files shipped with Titan. | 09 October 2014, 10:23:23 UTC |
| f56fa7e | Dan LaRocque | 09 October 2014, 10:17:35 UTC | Execute GotGFactory in a separate JVM In the aurelius-release profile, titan-dist-hadoop-2 automatically runs GraphOfTheGodsFactory with a BDB+ES config to generate database files for future compatibilty testing. This invocation tends to tip the Maven JVM over its memory limits when run inside the Maven JVM using exec:java. This commit changes it to exec:exec (an external JVM) to avoid memory pressure inside Maven. | 09 October 2014, 10:17:35 UTC |
| c1a4582 | Dan LaRocque | 09 October 2014, 10:16:47 UTC | Add section anchors inside titan-config-ref | 09 October 2014, 10:16:47 UTC |
| a8bad29 | Dan LaRocque | 09 October 2014, 10:03:55 UTC | Assembly & sample conf file templates refactoring This is for #743 and #762. This commit does a couple of related things. I could probably pull them into separate commits with some effort and additional testing to make sure the split was clean, but they've been gradually maturing in my git-stash for a while as a single changeset anyway. 1. The "filesystem" assembly descriptor, which copies files to the root of the repo (like bin/gremlin.sh) for testing during development, now just includes a component called "common" and nothing else. The "archive" assembly descriptor, which builds the zipfiles, also includes this component, plus a second for html/javadocs and a third for the readme-style text files in the root of the archive. This change eliminated the copy-paste duplication between the zipfile and root-of-dev-repo assemblies. The clean goal has also been updated to delete up rexhome and examples in the root of the repo, since the filesystem descriptor now includes those via the common component. This fixes #762. 2. The sample Titan conf files now pass through an additional preprocessor called ConfigurationFileFilter. This replaces lines of the form #TITANCFG{key=value} with a word-wrapped multi-line comment including information from the ConfigOption associated with key, including the description, default value, data type, and mutability. For mutabilities FIXED, GLOBAL_OFFLINE, and GLOBAL, there's also an tip to use the Management System API and reminder that changes to the value in the file have no effect after Titan is first started. After this comment appears the actual key=value pair on a line by itself. After ConfigurationFileFilter has completely processed the input file and written to a separate output file, it opens and reads the separate output file with ConfigurationLint. ConfigurationLint checks that the file parses as a PropertiesConfiguration and that every key is associated with a Titan ConfigOption. Furthermore, this commit extends ConfigurationLint to execute the verification predicate defined in each ConfigOption on its associated value in the separate output file. This serves as a sanity check on ConfigurationFileFilter. If ConfigurationFileFilter or ConfigurationLint find an error and exit with nonzero status, the build fails. This addresses #743. I could have probably figured out how to do all this with Velocity, but my aims were modest and I couldn't quickly figure out how to extend Velocity quite this way, so I just wrote this little utility program. | 09 October 2014, 10:03:55 UTC |
| 4c0be5e | Dan LaRocque | 07 October 2014, 18:08:12 UTC | Some minor typo fixes in Solr docs | 07 October 2014, 18:08:12 UTC |
| d2b037a | Dan LaRocque | 07 October 2014, 12:13:54 UTC | Make doc xref links same color as external links There was no real reason for xref links (internal to the documentation) to be different color from links to external resources. The xref color was also a very dark grey bordering on black, making them barely stand out as links. Now xref links are the same bright red as external links. | 07 October 2014, 12:13:54 UTC |
| 6f44e05 | Dan LaRocque | 07 October 2014, 12:08:20 UTC | Add Solr chapter to docs TOC | 07 October 2014, 12:08:20 UTC |
| c7ed2ae | Dan LaRocque | 07 October 2014, 12:04:08 UTC | Add a Solr doc section specific to DSE This commit fills the formerly-empty DSE Search section under the Solr doc chapter. The steps were tested on DSE 4.5.2 and a 0.5.1-SNAPSHOT version shortly before 0.5.1 went final. | 07 October 2014, 12:04:08 UTC |
| f741f8b | Dan LaRocque | 07 October 2014, 12:01:06 UTC | Add Solr troubleshooting entry on JTS jar Solr can throw an exception during core creation when the jts jar is absent from the Solr daemon classpath and schema.xml defines geo fields. This is a consequence of Spatial4j declaring JTS as an optional dependency. This commit adds a troubleshooting entry to the Solr docs that summarizes the resulting exceptions and describes determination of the appropriate JTS version to install. | 07 October 2014, 12:01:06 UTC |
| 3e44574 | Dan LaRocque | 07 October 2014, 11:59:57 UTC | Flesh out Solr dynamic fields docs a bit | 07 October 2014, 11:59:57 UTC |
| 1be6fa7 | Dan LaRocque | 07 October 2014, 11:58:18 UTC | Add sample config for Cassandra + Solr | 07 October 2014, 11:58:18 UTC |
| d816ef1 | Dan LaRocque | 07 October 2014, 10:12:36 UTC | Expand the manual Solr field def paragraph a bit | 07 October 2014, 10:12:36 UTC |
| bf6faa1 | Dan LaRocque | 07 October 2014, 10:12:05 UTC | Fix bogus copy-pasted section header in solr docs | 07 October 2014, 10:12:05 UTC |
| 6796f4d | Dan LaRocque | 07 October 2014, 10:08:29 UTC | Manage the spatial4j version in top-level pom Spatial4j is declared as a dependency in titan-core and in titan-solr. The version is now controlled from the top-level titan pom. This was necessary when compiling with an older, pre 4.8.1 Solr version that had a different Spatial4j dependency version. | 07 October 2014, 10:08:29 UTC |
| fa66b55 | Dan LaRocque | 07 October 2014, 10:07:29 UTC | Add titan-solr to titan-all Since titan-dist (which builds the zipfiles) relies on titan-all, this also adds titan-solr to the zipfiles. | 07 October 2014, 10:07:29 UTC |
| 7ea0946 | Matthias Broecheler | 07 October 2014, 02:37:17 UTC | Merge remote-tracking branch 'origin/titan05' into titan05 | 07 October 2014, 02:37:17 UTC |
| c6f96d9 | Matthias Broecheler | 07 October 2014, 02:37:05 UTC | Added solr documentation and expanded the description of some solr configuration options. | 07 October 2014, 02:37:05 UTC |
| c8a2321 | Dan LaRocque | 07 October 2014, 01:36:28 UTC | Fix getOrCreateVertexProperty param name This name change is cosmetic and does not affect the signature of the getOrCreateVertexProperty incremental loading method. Intent of this commit is just to make the code less confusing. | 07 October 2014, 01:36:28 UTC |
| 18bb53c | Dan LaRocque | 06 October 2014, 23:12:48 UTC | Merge branch 'titan05' of github.com:thinkaurelius/titan into titan05 | 06 October 2014, 23:12:48 UTC |
| 2881ed0 | Matthias Broecheler | 06 October 2014, 23:12:07 UTC | Added configuration namespace for solr | 06 October 2014, 23:12:07 UTC |
| 739e7d5 | Matthias Broecheler | 06 October 2014, 22:31:02 UTC | renamed core => collection throughout runner | 06 October 2014, 22:31:02 UTC |
| 866c90c | Matthias Broecheler | 06 October 2014, 22:29:27 UTC | renamed core => collection throughout index | 06 October 2014, 22:29:27 UTC |
| 3081cdf | Dan LaRocque | 06 October 2014, 22:22:47 UTC | Remove bogus comment from archive.xml descriptor This comment came with a copy-paste-edit but was never deleted/updated to reflect the local edits | 06 October 2014, 22:22:47 UTC |
| 47ee32a | Dan LaRocque | 06 October 2014, 22:22:20 UTC | Updated incremental loading sample script For #774 | 06 October 2014, 22:22:20 UTC |
| f6e44ef | Dan LaRocque | 06 October 2014, 20:14:13 UTC | Change context type in incr load method signature This is probably going to be a short-lived interface with TP3 around the corner, but the incremental loading script should probably use the more general TaskInputOutputContext interface as opposed to Mapper.Context or Reducer.Context, in case we want to call it from a Reducer at some point. TaskIOContext is the most general interface that still allows getCounter under both Hadoop 1 and 2. | 06 October 2014, 20:14:13 UTC |
| 477f043 | Dan LaRocque | 06 October 2014, 19:35:58 UTC | Updating Faunus incr. loading script imports The incremental loading script's set of default imports has changed: * removed "Mapper.Context" and added "Mapper", since we're using "Mapper.Context" in the documentation and example scripts * added TitanProperty | 06 October 2014, 19:35:58 UTC |
| 315ba6b | Dan LaRocque | 06 October 2014, 08:51:16 UTC | Add some debug statements to readHadoopVertex | 06 October 2014, 08:51:16 UTC |
| 3257577 | Dan LaRocque | 06 October 2014, 08:02:14 UTC | Fix convenience getters in FaunusSchemaManager FSM defines a general getRelation method, plus convenience methods for specific relation subtypes. FSM is just a pair of dumb ConcurrentHashMaps with no awareness of Titan schema. FSM.getRelation delegates to a helper class that knows about Titan's ManagementSystem API when a retrieval on one of the maps misses. However, the subtype helper getters did not delegate. This led to a confusing NPE in TitanHadoopGraph.readHadoopVertex's call to vertex.getTypeManager().getPropertyKey(name); the type was defined in FSM's helper class, but FSM was ignoring it. I think this was unintentional, since the contains methods on FSM also consult the ManagementSystem-aware helper. | 06 October 2014, 08:02:14 UTC |
| 07e2dfc | Dan LaRocque | 06 October 2014, 07:57:42 UTC | Suppress shutdown hook exception The code that stats an embedded Cassandra daemon for test purposes also installs a shutdown hook to kill the daemon. The hook calls a method which can also be called from a separate stop method by an ordinary thread outside the JVM shutdown hook sequence. This method attempts to uninstall the shutdown hook once it's done (useful on the stop path). However, uninstalling a shutdown hook when the JVM is trying to exit (the hook path) results in an IllegalStateException. It's scary looking but not harmful. This commit makes the method tolerate and log ISE. | 06 October 2014, 07:57:42 UTC |
| 21478eb | Matthias Broecheler | 03 October 2014, 18:04:16 UTC | Updated documentation | 03 October 2014, 18:04:16 UTC |
| bbd07f1 | Dan LaRocque | 03 October 2014, 11:31:27 UTC | Avoid Hadoop Job.toString The Job object's toString throws an exception depending on its internal state; it's unsafe to call it in the HBaseAuthHelper | 03 October 2014, 11:32:23 UTC |
| 9de9eb1 | Dan LaRocque | 03 October 2014, 10:49:34 UTC | Exclude some Cass RP/Murmur3 tests on HBase The tests excluded here failed on HBase because the tests expected that the backend would store keys without respect to their order. Should have been excluded on HBase from the start. | 03 October 2014, 10:49:34 UTC |
| f9101d4 | Dan LaRocque | 03 October 2014, 06:38:14 UTC | Remove UGI.getCredentials from HBase manager HBaseStoreManager contained some lines to dump delegation auth tokens, but these have compatibility problems between Hadoop 2 and 1. Removed. | 03 October 2014, 06:38:14 UTC |
| 316c1a1 | Dan LaRocque | 03 October 2014, 03:23:12 UTC | Merge branch 'lab41-solr' into titan05 Conflicts: titan-core/pom.xml titan-test/src/main/java/com/thinkaurelius/titan/graphdb/TitanIndexTest.java | 03 October 2014, 06:37:36 UTC |
| 06a3097 | Matthias Broecheler | 03 October 2014, 00:14:36 UTC | Merge branch 'titan05' of https://github.com/thinkaurelius/titan into titan05 Conflicts: titan-core/src/main/java/com/thinkaurelius/titan/graphdb/transaction/StandardTitanTx.java | 03 October 2014, 00:14:36 UTC |
| 9f02e0b | Matthias Broecheler | 03 October 2014, 00:03:49 UTC | Removed outdated code. | 03 October 2014, 00:03:49 UTC |
| 1ada4ef | Matthias Broecheler | 02 October 2014, 23:41:05 UTC | Implemented NOT and OR in Solr. Refactored query builder to use string instead of SolrQuery. | 02 October 2014, 23:41:05 UTC |
| 30c94dc | Matthias Broecheler | 02 October 2014, 22:52:17 UTC | Added test coverage for OR and NOT IndexQueries | 02 October 2014, 22:52:17 UTC |
| 80ac3f3 | Dan LaRocque | 02 October 2014, 22:18:26 UTC | Tool to generate config files from ConfigOptions Based on some feedback, I wanted a file templating facility that would let me both automatically annotate settings in the stock config files with ConfigOption information (like the description string and mutability level), but also do more pedestrian templating tasks, like including other files. The ConfigOption bit is probably possible with Velocity, but it seemed non-trivial or clumsy enough to implement with VTL that I just wrote a one-file tool to do the job instead. | 02 October 2014, 22:18:26 UTC |
| 8ab7fee | Matthias Broecheler | 02 October 2014, 22:17:43 UTC | Refactored SolrIndex code to make the tests work. Removed old mappings. Added dynamic field mappings to make schema handling easier. | 02 October 2014, 22:17:43 UTC |
| 1a7eda3 | Matthias Broecheler | 02 October 2014, 18:30:22 UTC | Added index.[*].map-name configuration option which makes it such that the field names are identical to the property key names in Titan. Added test cases. Addresses the first part of #770. | 02 October 2014, 18:30:22 UTC |
| af1cda2 | Matthias Broecheler | 02 October 2014, 17:36:38 UTC | Refactored and cleaned up SolrIndex, removed unnecessary config option, corrected field type mapping | 02 October 2014, 17:36:38 UTC |
| 35a4633 | Dan LaRocque | 02 October 2014, 08:56:56 UTC | Copy commented Titan-Hadoop config tips The stock config file conf/hadoop/titan-graphson.properties had a note about overridding Hadoop configuration settings along with some commented suggested Hadoop config key-value pairs to consider. This commit copies that chunk of comments to the other stock Titan-Hadoop config files. This should be templated one way or another so that we don't have to maintain a bunch of copies of the same chunk of text. | 02 October 2014, 08:56:56 UTC |
| 9518e73 | Dan LaRocque | 02 October 2014, 00:43:56 UTC | Prepping changelog for 0.5.1 This commit attempts to summarize changes between the 0.5.0 tag and titan05 as of 88a06b1045d0d3e2bc047a46b1002a50b5059fc1 (Oct 1 before the Solr merge). | 02 October 2014, 00:44:32 UTC |
| 88a06b1 | Dan LaRocque | 01 October 2014, 23:43:00 UTC | Tweak default cassandrathrift connection pool settings This commit disables the maxTotal parameter on the connection pool. The connection pool size is now regulated through the maxActive and maxIdle parameters on the pool. Setting maxTotal appears to cause commons-pool to repeatedly just barely exceed the limit by one connection, then close 15% of the total pool size before repeating the process. The overall number of connections stays at or just one over the configured limit, which is good, but routinely closing and reopening connections can lead to a tens of thousands of TIME_WAIT sockets in test scenarios with just one client and one server on the same physical host, which is bad. Keeping maxTotal off (negative) and using the active and idle limits instead seems to avoid this behavior. I've seen this before when testing Titan-Hadoop and worked around it there by overridding the defaults just for the affected Titan-Hadoop testsd. Now I'm seeing it again in ThriftGraphTest's getSlice stress tests, and I'm changing the defaults instead of just working around it in a specific test config. | 01 October 2014, 23:47:13 UTC |
| fc65db3 | Dan LaRocque | 01 October 2014, 23:40:31 UTC | Rewrote Titan-Hadoop incremental loading docs This feature has been restored as of bf39926f3f10ad8aacb16971d545c8595894e88c, but without documentation. Incremental loading is now documented in roughly the same place as it used to be in 0.4.4 (under the Titan I/O format docs), but the prose and code examples have been rewritten from scratch. | 01 October 2014, 23:42:40 UTC |
| d163815 | Dan LaRocque | 01 October 2014, 23:38:42 UTC | Adding concurrent getSlice stress KCVS tests I dislike nondeterministic tests like these in general, but they do have the potential to detect a hypothetical race condition that leads to a correctness failure. | 01 October 2014, 23:38:42 UTC |
| ca83628 | Pavel Yaskevich | 01 October 2014, 23:19:38 UTC | fix ThriftSolrTest to run in IDEA (+ name collision in testVertexTTLWithMixedIndices) | 01 October 2014, 23:19:38 UTC |
| c0b6e79 | Dan LaRocque | 01 October 2014, 19:48:08 UTC | Cassandra config tweak in Solr tests This is to support running the tests from non-Maven environments (in particular IDEA). | 01 October 2014, 19:48:08 UTC |
| ddb8af2 | Dan LaRocque | 01 October 2014, 19:29:37 UTC | Merge branch 'titan05' of github.com:thinkaurelius/titan into titan05 | 01 October 2014, 19:29:37 UTC |
| 69291d2 | Matthias Broecheler | 01 October 2014, 19:29:21 UTC | Changed field mapping in test to avoid conflict for solr adapter | 01 October 2014, 19:29:21 UTC |
| 38a4f12 | Pavel Yaskevich | 01 October 2014, 01:49:28 UTC | fix regex by removing filter from fieldType (+ handle empty facet/filter queries in AND statement correctly) | 01 October 2014, 01:49:28 UTC |
| 8889a15 | Matthias Broecheler | 01 October 2014, 00:52:39 UTC | Added method getVertices(long.. ids) to retrieve multiple vertices at once with better performance and usability than retrieving them individually. Fixes #575 | 01 October 2014, 00:52:39 UTC |
| 70f4a1b | Pavel Yaskevich | 30 September 2014, 23:59:33 UTC | replace filtered out text and fix titan.Decimal to map to double | 30 September 2014, 23:59:33 UTC |
| a5d4515 | Matthias Broecheler | 30 September 2014, 23:51:02 UTC | Renamed other exceptions to streamline exception handling. Calling setProperty for a multi-property now throws UnsupportedOperationException and out-of-transaction access to vertices throws IllegalStateException (both instead of IllegalARgumentException) | 30 September 2014, 23:51:02 UTC |
| 4c28718 | Matthias Broecheler | 30 September 2014, 23:23:11 UTC | Added SchemaViolationException and replaced current IllegalArgumentException. Added test case to verify. Fixes #681 | 30 September 2014, 23:23:11 UTC |
| 41abded | Matthias Broecheler | 30 September 2014, 22:36:16 UTC | Fixes #732 | 30 September 2014, 22:36:16 UTC |
| c304adb | Matthias Broecheler | 30 September 2014, 21:09:16 UTC | Added explicit field mapping to test case. | 30 September 2014, 21:09:16 UTC |
| f346aa9 | Dan LaRocque | 29 September 2014, 17:43:14 UTC | Kerberos+HBase+MapReduce authentication support This commit adds support for HBase's authentication token system, which is required for MapReduce (Titan-Hadoop/Faunus) jobs that want to talk to a Kerberized HBase instance. This commit has no effect on Kerberized HBase authentication outside of MapReduce/Titan-Hadoop/Faunus, which works as described in the HBase manual. This commit was tested on CentOS 5, RHEL 6, and MIT Kerberos 1.8.x. The Titan-Hadoop HBase read and write logic use different code and config keys. The read path uses HBase's own input format and config keys, while the write path uses a custom format brought in with Faunus. When writing without HBase configuration files (including hbase-site.xml) on the classpath of the MR tasks, the Titan-Hadoop config setting titan.hadoop.output.conf.ext.hbase.security.authentication should be set to Kerberos. This tells the HBaseStorageManager that starts up inside the Titan-Hadoop write path to check for a Kerberos authentication delegation token. If an hbase-site.xml containing the hbase.security.authentication=kerberos setting is on the MR task classpath, then there's no need to duplicate it in the Titan-Hadoop config file like this. The read path just needs that setting on the submitting process's classpath. | 30 September 2014, 08:35:44 UTC |
| cb7125e | Pavel Yaskevich | 30 September 2014, 07:24:16 UTC | fix integration tests startup (some of the tests still fail 6 out of 15) | 30 September 2014, 07:24:16 UTC |
| 4905804 | Matthias Broecheler | 30 September 2014, 00:54:08 UTC | Added comment on composite index with single key | 30 September 2014, 00:54:08 UTC |
| ba0736b | Pavel Yaskevich | 29 September 2014, 22:43:42 UTC | use core-template instead of individual solr cores, create configuration at runtime for tests | 29 September 2014, 22:43:42 UTC |
| dc46d5a | Matthias Broecheler | 29 September 2014, 19:42:26 UTC | Corrected test case and changed the semantics of Text.CONTAINS to AND instead of OR which is the Solr default | 29 September 2014, 19:42:26 UTC |
| 279a010 | Matthias Broecheler | 29 September 2014, 17:55:02 UTC | Added explicit exception when serializer returns null value in error as suggested by Praveen Peddi to make debugging easier. | 29 September 2014, 17:55:02 UTC |
| 6e3eaf5 | Matthias Broecheler | 27 September 2014, 21:42:11 UTC | Renamed getPropertyKey/getEdgeLabel to getOrCreate(PropertyKey/EdgeLabel) to provide a better description for the method semantics. Added getPropertyKey/EdgeLabel which only retrieves but does not attempt to create types (returns null if not exist). Added convenience methods containsPropertyKey/EdgeLabel. Refactored out schema related methods into SchemaInspector and SchemaManager. Refactored common graph access methods between TitanGraph and TitanTransaction into TitanGraphTransaction. Fixes #731. Fixes #767. | 27 September 2014, 21:42:11 UTC |
| f8c5c7b | Matthias Broecheler | 27 September 2014, 18:56:49 UTC | Refactored and extended index uniqueness tests | 27 September 2014, 18:56:49 UTC |
| 376bad2 | Matthias Broecheler | 26 September 2014, 20:45:06 UTC | Added test case and fix for stale vertex issue with multi query. Fixes #744. | 26 September 2014, 20:45:06 UTC |
| 4bc081b | Matthias Broecheler | 26 September 2014, 17:32:11 UTC | Added an exception type to distinguish backend failures caused by resource unavailability. In the management logger, swallow log exceptions due to an already closed log during shutdown. Fixes #726. | 26 September 2014, 17:32:11 UTC |
| 85f57c3 | Matthias Broecheler | 26 September 2014, 17:10:28 UTC | Added configuration option tx.recovery.verbose to enable std out messages for recovery. Fixes #707. | 26 September 2014, 17:10:28 UTC |
| e9c6a07 | Matthias Broecheler | 26 September 2014, 17:00:39 UTC | Lazy initialization of global configuraiton system in clopen. | 26 September 2014, 17:00:39 UTC |
| 23a463c | Dan LaRocque | 26 September 2014, 06:51:24 UTC | Merge branch 'titan05' of github.com:thinkaurelius/titan into titan05 | 26 September 2014, 06:51:24 UTC |
| bfc25a5 | Matthias Broecheler | 26 September 2014, 04:26:59 UTC | Marked the current instance with "(current)" and added check against closing the current instance. Fixes #708. | 26 September 2014, 04:26:59 UTC |
| cc1fc6a | Matthias Broecheler | 26 September 2014, 04:06:14 UTC | Fixes #766 | 26 September 2014, 04:06:14 UTC |
| e47bac7 | Matthias Broecheler | 26 September 2014, 03:44:01 UTC | Allow users to specify the element identifier used in IndexQueries so that special tokens can be used in case of a name clash (e.g. v.text:v.name). Added test coverage. | 26 September 2014, 03:44:01 UTC |
| 926b808 | Dan LaRocque | 26 September 2014, 01:56:28 UTC | Reduce illegal vertex ID log message severity Following comment by @mbroecheler on #727 | 26 September 2014, 01:56:28 UTC |
| fdab2e7 | Dan LaRocque | 25 September 2014, 22:30:17 UTC | Revert ANTLR 3.x from 3.5 to 3.2 This should fix Cassandra (which compiled against 3.2 and depends on the org.antlr.runtime.Tokens.EOF_TOKEN field removed in 3.3) but could have broken Solr, which currently depends on ANTLR 3.5. Can't tell since the Solr have preexisting config problems and don't pass right now. If Solr can't use ANTLR 3.2, then we'll may have to use the shade plugin to ship multiple copies of ANTLR in order to allow Cassandra and Solr to coexist in the same VM. | 25 September 2014, 22:30:17 UTC |
| d4056cb | Matthias Broecheler | 25 September 2014, 18:18:03 UTC | Finised streamlining test cases and query semantics across the different indexing backends. Added additional test coverage for Lucene. SOLR does not yet work. | 25 September 2014, 18:18:03 UTC |
| 443428a | Dan LaRocque | 25 September 2014, 17:37:10 UTC | Dependency management tweaks for BP 2.6.0 Titan is still on BP 2.5.0 after this commit. This commit just adds some dependencyManagement bits to the poms to allow changing the version to 2.6.0 for testing without causing convergence errors. | 25 September 2014, 17:37:10 UTC |
| 0c4e35a | Dan LaRocque | 25 September 2014, 17:19:22 UTC | Enable BP supportsThreadIsolatedTransactions Reflectively lookup and set this Blueprints Features field to true when it exists, otherwise log and fail silently. This is done reflectively for transparent compatibility with 2.5.0 and 2.6.0 without having to recompile (it exists in the latter but not in the former). | 25 September 2014, 17:19:22 UTC |
| 5f56276 | Dan LaRocque | 25 September 2014, 16:27:05 UTC | Always close IndexReaders in LuceneIndex.mutate For #761 | 25 September 2014, 16:27:05 UTC |
| 6c3914a | Dan LaRocque | 25 September 2014, 16:25:06 UTC | Update partitioned-vertex-label tests This is a consequence of 628d48fe23431422e71225306ce2220611a45c86. One of the test methods in TitanBaseGraphTest creates a partitioned vertex label, but the ability to create these is now dependendent on whether cluster.partition is true. I made the label in TitanBaseGraphTest non-partitioned moved the partitioned-vertex-label related assertions to a smaller method in TitanPartitionGraphTest. | 25 September 2014, 16:25:06 UTC |
| d787396 | Dan LaRocque | 24 September 2014, 21:56:35 UTC | Warn and return null for illegal getVertex(id) arg Since this method is part of the Blueprints API, it should probably return null instead of throwing IllegalArgumentException for the (Titan-specific rather than Blueprints-standardized) range of IDs which are outside the legal vertex ID space. This commit changes the exception type thrown in IDManager.getUserVertexIDType to InvalidIDException, which inherits from TitanException and in turn RuntimeException. StandardTitanTx.getVertex catches InvalidIDException, logs a warning, and then returns null. I think nothing needs to be done to make bogus schema vertex IDs return null, since they follow a different logic path in IDManager.getKey that just converts the long into a byte buffer. Not sure though. | 24 September 2014, 22:00:56 UTC |
| 628d48f | Dan LaRocque | 24 September 2014, 17:42:52 UTC | Removing cluster.partition check in tx.addVertex Part of #706; Preceding commit 6941008da884bedde58dd516d19e5090c55f7811 has moved this check up to schema definition time, making the check inside addVertex mostly redundant (and entirely redundant on new installs). | 24 September 2014, 22:00:56 UTC |
| 024c667 | Dan LaRocque | 24 September 2014, 17:40:13 UTC | Check cluster.partition in part vlabel creation VertexLabelMaker now checks that cluster.partition is true in the graph configuration when it creates a label with `partition()` enabled. Part of #706. This makes the related cluster.partition check in `StandardTitanTx.addVertex` redundant for new installations, though 0.5.0 installs could have defined a partitioned vertex label on a non cluster.partition'ed backend. | 24 September 2014, 22:00:47 UTC |
| 899c428 | Matthias Broecheler | 23 September 2014, 18:44:45 UTC | Intermediate index related refactoring to streamlining semantics and index backend interaction. | 23 September 2014, 18:44:45 UTC |
| 2621fa6 | Dan LaRocque | 23 September 2014, 15:39:29 UTC | Update reference to old "autotype" setting Should be schema.default | 23 September 2014, 15:39:29 UTC |
| 0bf7860 | Dan LaRocque | 19 September 2014, 05:59:15 UTC | Make ES tests log on the console Cheating a bit to make debugging a test setup failure on CI easier | 19 September 2014, 05:59:23 UTC |
| d5f810b | Dan LaRocque | 19 September 2014, 05:56:51 UTC | Fix InMemoryPartitionGraphTest I broke this in 1e5e54bf5cfad4aa4c4540887c9ab8f07e933d1f when adding a mix of partitioned and unpartitioned test methods to its superclass; I neglected InMemory entirely while thinking about Cassandra and HBase subclasses which have the same superclass. | 19 September 2014, 05:56:51 UTC |
| 166730e | Dan LaRocque | 19 September 2014, 04:24:24 UTC | Rename ALLOW_STALE_CONFIG option string For #743 | 19 September 2014, 04:26:06 UTC |
