Hey, I personally haven't had a chance to test M9 at this stage as our system depends on titan and I've been waiting for an M9 enabled titan (well....It depends on gremlin-server but we have quite a bit of titan specific code at the moment and I haven't had time to switch things around for testing).
Our last tests were on titan 0.9-M1 (TP3 M6) and failed to pass our initial tests because of the id lossiness issues that have since been corrected for M9 so I don't know what will come of the other tests. We make heavy use of Gremlin in a non-java environment so our feedback will probably mostly relate to gremlin-server. It really depends on whether or not you think our feedback is likely to bring breaking changes. I don't see a reason why though? With all that said and done, if anyone can provide me with an M9 enabled titan that would make my day and feedback should follow relatively quickly. On Wed, May 20, 2015 at 1:28 PM, Marko Rodriguez <[email protected]> wrote: > Hello TinkerPop vendors, > > Please see: > > > https://git-wip-us.apache.org/repos/asf?p=incubator-tinkerpop.git;a=tag;h=refs/tags/3.0.0.M9-incubating-rc3 > > If vendors could test against their products that would be excellent. We > are getting very very close. I've been working closely with the Titan > codebase and everything is pretty spick and span on that front and they do > a lot in terms of vendor-specific steps, traversal strategies, OLAP > GraphComputer, transactions, etc. Please report back your requirements. > > Here our our last two tickets for GA: > > https://issues.apache.org/jira/browse/TINKERPOP3-549?jql=fixVersion%20%3D%203.0.0.GA%20AND%20project%20%3D%20TINKERPOP3%20AND%20resolution%20%3D%20Unresolved%20ORDER%20BY%20priority%20DESC > However, if people are interested in an M9, we are more than happy to do > so, so please express your desires. > > The CHANGELOG since RC2 is: > > * TinkerGraph classes are now final to restrict user and vendor extension. > * Added `TraversalStrategy.VendorOptimization` to ensure that all > TinkerPop optimizations execute first on the known TinkerPop steps. > * Added `TailGlobalStep` and `TailLocalStep` (`tail()`) which gets objects > from the end of the traversal stream. > * `AndStep` and `OrStep` are now simply markers where > `WhereStep(a.and(b).and(c)...and(z))` is the compilation. > * Moved `Compare`, `Contains`, `Order`, `Operator`, and `P` to > `process/traversal` from `structure/` as they are process-based objects. > * `HasContainer` now uses `P` predicate as helper methods and tests are > more thorough on `P`. > * Changed Gremlin Server integration/performance tests to be runnable from > within the `gremlin-server` directory or from the project root. > * Moved the string methods of `TraversalHelper` to `StringFactory`. > * Renamed JSON-related serializers for Gremlin Server to be more > consistent with GraphSON naming. > * Removed `HasTraversalStep` in favor of new `P.traversal` model with > `HasStep`. > * Fixed bug in `WsGremlinTextRequestDecoder` where custom serializers from > graphs were not being used. > * Added `AndP` which allows for the `and()`-ing of `P` predicates. > * `Order.opposite()` is now `reversed()` as that is a `Comparator` > interface method with the same semantics. > * `Compare/Contains/P.opposite()` are now `negate()` as that is a > `BiPredicate` interface method with the same semantics. > * `has(traversal)` is replaced by `where(traversal)` and > `has(key,traversal)`. `HasXXX` is always with respects to an element > property. > * Added `TraversalScriptHelper` with static methods for dynamically > creating a `Traversal` from a JSR 223 `ScriptEngine`. > * Changed `SubgraphStrategy` to take `Traversal` rather than `Predicate` > for filtering. > * Improved `SubgraphStrategy` to only modify the `Traversal` if filtering > was required. > * Improved logging of errors in the `HttpGremlinEndpointHandler` to > include a stracktrace if one was present. > * Moved `AbstractGremlinSuite.GraphProviderClass` to > `org.apache.tinkerpop.gremlin.GraphProviderClass`. > * Simplified the Gremlin-Groovy test suite where there is now no > distinction between `STANDARD` and `COMPUTER` tests. > * `VertexProgram` and `MapReduce` now add a `Graph` parameter to > `loadState(Graph, Configuration)`. > * Added `ScopingStrategy` which auto-scopes `select()` and `where()` so > the language looks clean. > * Added `Scoping` as a marker interface to state that a step desires a > particular `Scope`. > * `SelectStep`, `SelectOneStep`, and `WhereStep` support both > `Scope.local` and `Scope.global` for `Map<String,Object>` or `Path` > analysis, respectively. > * Fixed a bug in the `TraversalStrategies` sort algorithm. > * Removed numerous unused static utility methods in `TraversalHelper`. > * TinkerGraph process suite tests are now running with and without > strategies in place. > * Added `IncidentToAdjacentStrategy` which rewrites `outE().inV()`, > `inE().outV()` and `bothE().otherV()` to `out()`, `in()` and `both()` > respectively. > * Renamed `ComparatorHolderRemovalStrategy` to > `OrderGlobalRemovalStrategy` as it now only applies to `OrderGlobalStep`. > * Anonymous traversal no longer have `EmptyGraph` as their graph, but > instead use `Optional<Graph>.isPresent() == false`. > * Added `Traversal.Admin.setGraph(Graph)` as strategies that need > reference to the graph, need it across all nested traversals. > * `AbstractLambdaTraversal` is now smart about `TraversalParent` and > `TraversalStrategies`. > > Take care, > Marko. > > http://markorodriguez.com > >
