tinkerpop-commits mailing list archives: July 2016

Site index · List index
Message list1 · 2 · 3 · 4 · 5 · 6 · Next »Thread · Author · Date
dkupp...@apache.org tinkerpop git commit: TINKERPOP-1355 Design HasContainer for extension Wed, 06 Jul, 11:05
dkupp...@apache.org tinkerpop git commit: Install Python setuptools prior each build in Docker as it's now required by a maven build step. Fri, 08 Jul, 21:12
ok...@apache.org [1/5] tinkerpop git commit: When a traversal is being constructed its construction history is maintained in a new ByteCode object. ByteCode is simply a list of instructions (stepName/arguments) that denote exactly how the traversal was constructed. This Fri, 01 Jul, 22:23
ok...@apache.org [2/5] tinkerpop git commit: When a traversal is being constructed its construction history is maintained in a new ByteCode object. ByteCode is simply a list of instructions (stepName/arguments) that denote exactly how the traversal was constructed. This Fri, 01 Jul, 22:23
ok...@apache.org [3/5] tinkerpop git commit: When a traversal is being constructed its construction history is maintained in a new ByteCode object. ByteCode is simply a list of instructions (stepName/arguments) that denote exactly how the traversal was constructed. This Fri, 01 Jul, 22:23
ok...@apache.org [4/5] tinkerpop git commit: When a traversal is being constructed its construction history is maintained in a new ByteCode object. ByteCode is simply a list of instructions (stepName/arguments) that denote exactly how the traversal was constructed. This Fri, 01 Jul, 22:23
ok...@apache.org [5/5] tinkerpop git commit: When a traversal is being constructed its construction history is maintained in a new ByteCode object. ByteCode is simply a list of instructions (stepName/arguments) that denote exactly how the traversal was constructed. This Fri, 01 Jul, 22:23
ok...@apache.org [1/2] tinkerpop git commit: TINKERPOP-1341 Add missing classes to GryoRegistrator Sat, 02 Jul, 13:31
ok...@apache.org [2/2] tinkerpop git commit: Merge branch 'TINKERPOP-1341' of https://github.com/dalaro/incubator-tinkerpop Sat, 02 Jul, 13:31
ok...@apache.org [01/11] tinkerpop git commit: Improved performance of TinkerGraph around element lookup. Sat, 02 Jul, 13:40
ok...@apache.org [02/11] tinkerpop git commit: Made TinkerGraph validate that ids be homogeneous up front. Sat, 02 Jul, 13:40
ok...@apache.org [03/11] tinkerpop git commit: Updated changelog. Sat, 02 Jul, 13:40
ok...@apache.org [04/11] tinkerpop git commit: removed NOTE: Neo4jGraph currently does not support HA and added an image for the HA section. CTR. Sat, 02 Jul, 13:40
ok...@apache.org [05/11] tinkerpop git commit: Cancellation of a timeout should occur in existing thread pool Sat, 02 Jul, 13:40
ok...@apache.org [06/11] tinkerpop git commit: TINKERPOP-1341 Add missing classes to GryoRegistrator Sat, 02 Jul, 13:40
ok...@apache.org [07/11] tinkerpop git commit: Removed potential for NPEs when validating IDs given to vertices()/edges() Sat, 02 Jul, 13:40
ok...@apache.org [08/11] tinkerpop git commit: Merge remote-tracking branch 'origin/TINKERPOP-1063' into tp31 Sat, 02 Jul, 13:40
ok...@apache.org [09/11] tinkerpop git commit: Merge remote-tracking branch 'origin/tp31' Sat, 02 Jul, 13:40
ok...@apache.org [10/11] tinkerpop git commit: Merge branch 'TINKERPOP-1341' of https://github.com/dalaro/incubator-tinkerpop Sat, 02 Jul, 13:40
ok...@apache.org [11/11] tinkerpop git commit: Merge branch 'master' into TINKERPOP-1278 Sat, 02 Jul, 13:40
ok...@apache.org tinkerpop git commit: added toString(), equals(), and hashCode() to ByteCode and Instruction. Sat, 02 Jul, 17:28
ok...@apache.org tinkerpop git commit: fixed ByteCode.addStep() bug in select() step. Tue, 05 Jul, 14:22
ok...@apache.org tinkerpop git commit: Created ReflectionTranslator which uses Java reflection to generate a Traversal from ByteCode in Java. If ByteCode does not have lambdas, then translators can default to this translator when evaluating on the JVM. This is all settin Tue, 05 Jul, 18:13
ok...@apache.org tinkerpop git commit: I think I have the Translator API finalized -- its simple: source, anonymous, translate(ByteCode). That is, every translator needs to know the source and anonmyous traversal classes and from there, the ByteCode is used to populate a Tue, 05 Jul, 22:09
ok...@apache.org tinkerpop git commit: added a method cache so looking up methods via reflection is fast. Wed, 06 Jul, 14:00
ok...@apache.org tinkerpop git commit: Void and void can not use FieldSerializers. However, note that Void and void are in DefaultSerializers and thus, don't need to be registered by GryoRegistrator. Commented them out and added a note if for some reason Void and void re Wed, 06 Jul, 14:27
ok...@apache.org [1/3] tinkerpop git commit: TINKERPOP-1355 Design HasContainer for extension Wed, 06 Jul, 14:46
ok...@apache.org [2/3] tinkerpop git commit: Void and void can not use FieldSerializers. However, note that Void and void are in DefaultSerializers and thus, don't need to be registered by GryoRegistrator. Commented them out and added a note if for some reason Void and v Wed, 06 Jul, 14:46
ok...@apache.org [3/3] tinkerpop git commit: Merge branch 'master' into TINKERPOP-1278 Wed, 06 Jul, 14:46
ok...@apache.org [1/3] tinkerpop git commit: renamed ByteCode to bytecode. Added Python-version of Bytecode to gremlin-python. Updated accordingly. Next up, Python-based RemoteGraph, TraversalStrategies, RemoteStrategy, and TranslationStrategy. From there, gremlin-python Wed, 06 Jul, 16:02
ok...@apache.org [2/3] tinkerpop git commit: renamed ByteCode to bytecode. Added Python-version of Bytecode to gremlin-python. Updated accordingly. Next up, Python-based RemoteGraph, TraversalStrategies, RemoteStrategy, and TranslationStrategy. From there, gremlin-python Wed, 06 Jul, 16:02
ok...@apache.org [3/3] tinkerpop git commit: renamed ByteCode to bytecode. Added Python-version of Bytecode to gremlin-python. Updated accordingly. Next up, Python-based RemoteGraph, TraversalStrategies, RemoteStrategy, and TranslationStrategy. From there, gremlin-python Wed, 06 Jul, 16:02
ok...@apache.org tinkerpop git commit: Okay, here it is -- Python versions of Graph, RemoteGraph, TraversalStrategies, TraversalStrategy. All connected up and working. You no longer create a PythonGraphTraversal, but instead a RemoteGraph and then do graph.traversal() ju Wed, 06 Jul, 16:57
ok...@apache.org tinkerpop git commit: PythonGraphTraversal is now GraphTraversal, likewise for PythonTraversal and PythonGraphTraversalSource. Bytecode (in python) handles bindings correctly. Added a test to RESTRemoteConnectionTest to ensure proper use of bindings. Org Wed, 06 Jul, 19:12
ok...@apache.org tinkerpop git commit: various cleanups and organizations. added nice __repr__s in Python that mirror the Java toStrings(). Tweaked with the play code that generates JSON from Bytecode -- no need to have nested lists for arguments as the head of the list Wed, 06 Jul, 20:54
ok...@apache.org tinkerpop git commit: found a bug in how bindings were being handled in Gremlin-Python -- fixed and added test case to verify it. Simplified Gremlin-Python Bytecode implementation a bit. Wed, 06 Jul, 21:55
ok...@apache.org [01/13] tinkerpop git commit: Fixed a problem with how timeouts were scheduled in GremlinExecutor Wed, 06 Jul, 22:19
ok...@apache.org [02/13] tinkerpop git commit: Added more invalid binding keys to Gremlin Server OpProcessor validation. Wed, 06 Jul, 22:19
ok...@apache.org [03/13] tinkerpop git commit: ResultSet.hasNext should be idempotent Wed, 06 Jul, 22:19
ok...@apache.org [04/13] tinkerpop git commit: Update changelog. Wed, 06 Jul, 22:19
ok...@apache.org [05/13] tinkerpop git commit: Update changelog. Wed, 06 Jul, 22:19
ok...@apache.org [06/13] tinkerpop git commit: ResultSet.hasNext should be idempotent Wed, 06 Jul, 22:19
ok...@apache.org [07/13] tinkerpop git commit: This closes #346 Wed, 06 Jul, 22:19
ok...@apache.org [08/13] tinkerpop git commit: Merge remote-tracking branch 'origin/TINKERPOP-1350' into tp31 Wed, 06 Jul, 22:19
ok...@apache.org [09/13] tinkerpop git commit: Merge remote-tracking branch 'origin/tp31' Wed, 06 Jul, 22:19
ok...@apache.org [10/13] tinkerpop git commit: Merge remote-tracking branch 'origin/TINKERPOP-1354' into tp31 Wed, 06 Jul, 22:19
ok...@apache.org [11/13] tinkerpop git commit: Merge remote-tracking branch 'origin/tp31' Wed, 06 Jul, 22:19
ok...@apache.org [12/13] tinkerpop git commit: Fixed merge issues from previous commit. Wed, 06 Jul, 22:19
ok...@apache.org [13/13] tinkerpop git commit: Merge branch 'master' into TINKERPOP-1278 Wed, 06 Jul, 22:19
ok...@apache.org tinkerpop git commit: added ProcessStandard and ProcessComputer tests for JavaTranslator which basically generates a traversal bytecode, then clears the traversal of its steps and uses JavaTranslator to regenerate the traversal steps. Was in a reflection Thu, 07 Jul, 12:46
ok...@apache.org tinkerpop git commit: subinterfaced Translator with ScriptTranslator and StepTranslator. Basically, any translator that is oriented towards a ScriptEngine is a ScriptTranslator. Any translator that generates 'native' step code is a StepTranslator. Cleane Thu, 07 Jul, 14:53
ok...@apache.org tinkerpop git commit: learned more about modules vs. packages in Python. Combined classes into a single .py file. Things are much cleaner now... all the traversal stuff is in traversal.py (strategies, strategy, translator, bytecode, traversal, traversal Thu, 07 Jul, 15:52
ok...@apache.org [1/3] tinkerpop git commit: reviewed @davebshow old PR and repackaged Gremlin-Python accordingly. gremlin_python/driver, structure, process. Thats that.' Thu, 07 Jul, 18:17
ok...@apache.org [2/3] tinkerpop git commit: reviewed @davebshow old PR and repackaged Gremlin-Python accordingly. gremlin_python/driver, structure, process. Thats that.' Thu, 07 Jul, 18:17
ok...@apache.org [3/3] tinkerpop git commit: reviewed @davebshow old PR and repackaged Gremlin-Python accordingly. gremlin_python/driver, structure, process. Thats that.' Thu, 07 Jul, 18:17
ok...@apache.org tinkerpop git commit: couldn't get native Python lambda code instrospection to work generally. suxoid....... howevewr, I did do some pom.xml cleanup and META-INF shizzzzz. Thu, 07 Jul, 19:58
ok...@apache.org tinkerpop git commit: while reviewing @twilmes work, I made various adjustments. Coding style things like finalizing variables, simplifiying some methods using existing Helper classes we have, added more tests to PrunePathStrategyTest, added PathProcess Fri, 08 Jul, 16:01
ok...@apache.org tinkerpop git commit: found a bug in TraversalVertexProgramStep where it was not respecting the traversals registered strategies. Fri, 08 Jul, 19:13
ok...@apache.org tinkerpop git commit: ImmutablePath.Head can be retrated -- it simply returns itself as it has no data. Removed a System.out.println I left in PrunePathStrategyTest. Tweaking MatchStep slightly around tags -- found a few archaic lines that were pointless Fri, 08 Jul, 20:19
ok...@apache.org tinkerpop git commit: Added OLTP optimization where a barrier is added after a path processor to try and thin the stream. When we move to having LazyBarrierStrategy go prime time, we should move that code to there. Added more test cases to PrunePathStrat Fri, 08 Jul, 22:29
ok...@apache.org tinkerpop git commit: Fixed an NPE around path and nested children with path processors. Added some TODO notes in PrunePathStrategyTest about why keepLabels are the way they are -- I believe that nesting is not sound and should be looked into more. And m Fri, 08 Jul, 22:58
ok...@apache.org tinkerpop git commit: made PathPruneStategyTest easier to write/read by using the toString() of the list nesting. Fri, 08 Jul, 23:11
ok...@apache.org tinkerpop git commit: added more PrunePathStrategyTest patterns and more TODO notes. So the current model works great for nested repeat() as with repeat, there is recurssion on the child. However, for non-looping nests, bad. Sat, 09 Jul, 00:35
ok...@apache.org tinkerpop git commit: a test failed in TinkerGraphNoStrategyComputerIntegrateTest. It must have been wrong for a long time but we just got 'thread lucky.' FilterRankStrategy is required or else the ordering goes bad after dedup(). Sat, 09 Jul, 13:25
ok...@apache.org tinkerpop git commit: added more test cases and have 2 double nested traversals that are broken. Sat, 09 Jul, 14:58
ok...@apache.org tinkerpop git commit: added a test around paths where once past the path, back to selective processing. Sat, 09 Jul, 15:47
ok...@apache.org tinkerpop git commit: added a lazy barrier to OLTP MatchStep (added a TODO note that in the future we may want to make MatchStep a LocalBarrierStep). Tweaked up the OLTP PrunePathStrategy optimization to only do lazy barriers if the parent is NOT a Match Sun, 10 Jul, 18:48
ok...@apache.org tinkerpop git commit: renamed PrunePathStrategy to PathRetractionStrategy to better align with the Path API retract() naming. PathRetractionStrategy should NOT execute if there are lambdas steps or path steps. Do a global analysis to ensure that. Added a Mon, 11 Jul, 15:49
ok...@apache.org tinkerpop git commit: added more test patterns to PathRestractionStrategy ... tweaked up NativeNeo4jCypherCheck as not all results need to return everything matched. Mon, 11 Jul, 16:46
ok...@apache.org tinkerpop git commit: using memoization in MatchEndStep so referenced labels don't need to be computed over and over again. Simplified PathUtil -- removed two methods that are now inlined. Mon, 11 Jul, 18:11
ok...@apache.org tinkerpop git commit: Greatly simplified ImmutablePath and MutablePath.retract() algorithms. Got a 3x speedup on a big ol fat match() query I'm benchmarking with. Simplified LP_XXXX_Traverser implementations around retract(). Small fix to NativeNeo4jCyph Mon, 11 Jul, 19:09
ok...@apache.org tinkerpop git commit: optimized Path.equals() in both ImmutablePath and MutablePath.. Lots of unnecessary object creation removed. Now that path equality is likely with bulking due to Path retraction, having equals() fast and clean is important. Mon, 11 Jul, 19:33
ok...@apache.org [01/22] tinkerpop git commit: TINKERPOP-1254 Support dropping traverser path information when it is no longer needed Mon, 11 Jul, 19:51
ok...@apache.org [02/22] tinkerpop git commit: while reviewing @twilmes work, I made various adjustments. Coding style things like finalizing variables, simplifiying some methods using existing Helper classes we have, added more tests to PrunePathStrategyTest, added Path Mon, 11 Jul, 19:51
ok...@apache.org [03/22] tinkerpop git commit: found a bug in TraversalVertexProgramStep where it was not respecting the traversals registered strategies. Mon, 11 Jul, 19:51
ok...@apache.org [04/22] tinkerpop git commit: ImmutablePath.Head can be retrated -- it simply returns itself as it has no data. Removed a System.out.println I left in PrunePathStrategyTest. Tweaking MatchStep slightly around tags -- found a few archaic lines that were p Mon, 11 Jul, 19:51
ok...@apache.org [05/22] tinkerpop git commit: Added OLTP optimization where a barrier is added after a path processor to try and thin the stream. When we move to having LazyBarrierStrategy go prime time, we should move that code to there. Added more test cases to PruneP Mon, 11 Jul, 19:51
ok...@apache.org [06/22] tinkerpop git commit: Fixed an NPE around path and nested children with path processors. Added some TODO notes in PrunePathStrategyTest about why keepLabels are the way they are -- I believe that nesting is not sound and should be looked into mor Mon, 11 Jul, 19:51
ok...@apache.org [07/22] tinkerpop git commit: made PathPruneStategyTest easier to write/read by using the toString() of the list nesting. Mon, 11 Jul, 19:51
ok...@apache.org [08/22] tinkerpop git commit: added more PrunePathStrategyTest patterns and more TODO notes. So the current model works great for nested repeat() as with repeat, there is recurssion on the child. However, for non-looping nests, bad. Mon, 11 Jul, 19:51
ok...@apache.org [09/22] tinkerpop git commit: a test failed in TinkerGraphNoStrategyComputerIntegrateTest. It must have been wrong for a long time but we just got 'thread lucky.' FilterRankStrategy is required or else the ordering goes bad after dedup(). Mon, 11 Jul, 19:51
ok...@apache.org [10/22] tinkerpop git commit: added more test cases and have 2 double nested traversals that are broken. Mon, 11 Jul, 19:51
ok...@apache.org [11/22] tinkerpop git commit: added a test around paths where once past the path, back to selective processing. Mon, 11 Jul, 19:51
ok...@apache.org [12/22] tinkerpop git commit: Updating strategy to fix nesting issues and moved label dropping from MatchStep's processNextStart to MatchEndStep's processNextStart. Mon, 11 Jul, 19:51
ok...@apache.org [13/22] tinkerpop git commit: Merge branch 'TINKERPOP-1254' of https://git-wip-us.apache.org/repos/asf/tinkerpop into TINKERPOP-1254 Mon, 11 Jul, 19:51
ok...@apache.org [14/22] tinkerpop git commit: Updated ReferencePath to copy labels into a new HashSet before adding them to reference path so that ReferencePaths will not end up sharing label sets. Added a number of new PrunePathStrategy tests. Mon, 11 Jul, 19:51
ok...@apache.org [15/22] tinkerpop git commit: added a lazy barrier to OLTP MatchStep (added a TODO note that in the future we may want to make MatchStep a LocalBarrierStep). Tweaked up the OLTP PrunePathStrategy optimization to only do lazy barriers if the parent is NOT Mon, 11 Jul, 19:51
ok...@apache.org [16/22] tinkerpop git commit: renamed PrunePathStrategy to PathRetractionStrategy to better align with the Path API retract() naming. PathRetractionStrategy should NOT execute if there are lambdas steps or path steps. Do a global analysis to ensure that. Mon, 11 Jul, 19:51
ok...@apache.org [17/22] tinkerpop git commit: added more test patterns to PathRestractionStrategy ... tweaked up NativeNeo4jCypherCheck as not all results need to return everything matched. Mon, 11 Jul, 19:51
ok...@apache.org [18/22] tinkerpop git commit: using memoization in MatchEndStep so referenced labels don't need to be computed over and over again. Simplified PathUtil -- removed two methods that are now inlined. Mon, 11 Jul, 19:51
ok...@apache.org [19/22] tinkerpop git commit: Greatly simplified ImmutablePath and MutablePath.retract() algorithms. Got a 3x speedup on a big ol fat match() query I'm benchmarking with. Simplified LP_XXXX_Traverser implementations around retract(). Small fix to NativeN Mon, 11 Jul, 19:51
ok...@apache.org [20/22] tinkerpop git commit: optimized Path.equals() in both ImmutablePath and MutablePath.. Lots of unnecessary object creation removed. Now that path equality is likely with bulking due to Path retraction, having equals() fast and clean is important. Mon, 11 Jul, 19:51
ok...@apache.org [21/22] tinkerpop git commit: Merge branch 'TINKERPOP-1254' Mon, 11 Jul, 19:51
ok...@apache.org [22/22] tinkerpop git commit: updated CHANGELOG with PathRetractionStrategy work by @twilmes. Mon, 11 Jul, 19:51
ok...@apache.org tinkerpop git commit: added optimized versions of labels() and objects() for ImmutablePath and added more test cases to PathTest to ensure the path retraction is sound. It is. CTR. Mon, 11 Jul, 22:55
ok...@apache.org tinkerpop git commit: Found a bug in the PathRetractStrategy that occurs when the match() select() keepLabels do not include the match dedupLabels that were backpropagated into the MatchStep via MatchPredicateStrategy. This caused a 'label could not be f Tue, 12 Jul, 13:47
ok...@apache.org [1/2] tinkerpop git commit: realized that a few traverser species were not super.merge() and thus, not having their tags merged appropriately. this led to weird results in MatchStep when prune strategy was activated. However, this then exposed a massive Tue, 12 Jul, 17:53
ok...@apache.org [2/2] tinkerpop git commit: a better generator for referenced labels in MatchStep related to PathRetractionStrategy. One more issue I found is coming up next and then I'm moving on from this. CTR. Tue, 12 Jul, 17:53
ok...@apache.org tinkerpop git commit: found a happy medium -- if MatchStep.keepLabels contains both the matchStartLabels and matchEndLabels, then don't prune as it does nothing but incur runtime. This is a clean solution that ensures that full select()s are not pointles Tue, 12 Jul, 20:22
ok...@apache.org tinkerpop git commit: a bunch of minor optimizations to the Traverser species (direct refernece calls instead of method calls) and made MatchStep a bit more efficient too for local barrier. Tue, 12 Jul, 23:20
Message list1 · 2 · 3 · 4 · 5 · 6 · Next »Thread · Author · Date
Box list
Apr 2021201
Mar 2021172
Feb 2021152
Jan 2021127
Dec 2020186
Nov 2020200
Oct 2020306
Sep 2020227
Aug 2020159
Jul 202074
Jun 2020214
May 2020176
Apr 2020266
Mar 2020293
Feb 2020257
Jan 2020304
Dec 2019299
Nov 2019288
Oct 2019277
Sep 2019431
Aug 2019281
Jul 2019467
Jun 2019574
May 2019342
Apr 2019240
Mar 2019448
Feb 2019282
Jan 2019487
Dec 2018433
Nov 2018459
Oct 2018943
Sep 2018516
Aug 2018960
Jul 2018605
Jun 2018845
May 2018866
Apr 20181008
Mar 20181185
Feb 2018379
Jan 2018723
Dec 2017333
Nov 2017779
Oct 2017473
Sep 2017884
Aug 2017494
Jul 2017992
Jun 20171082
May 2017660
Apr 2017337
Mar 2017729
Feb 20171052
Jan 20171626
Dec 2016869
Nov 20161023
Oct 20161010
Sep 2016910
Aug 2016598
Jul 2016537
Jun 2016787
May 2016660
Apr 2016422
Mar 20161169
Feb 2016852
Jan 2016316
Dec 2015523
Nov 2015485
Oct 2015838
Sep 2015535
Aug 2015322
Jul 2015480
Jun 2015758
May 2015909
Apr 2015954
Mar 2015613
Feb 2015351
Jan 20159