Package org.apache.tinkerpop.gremlin.process.traversal.strategy.optimization

Class IdentityRemovalStrategy

java.lang.Object
org.apache.tinkerpop.gremlin.process.traversal.strategy.AbstractTraversalStrategy<TraversalStrategy.OptimizationStrategy>
org.apache.tinkerpop.gremlin.process.traversal.strategy.optimization.IdentityRemovalStrategy
All Implemented Interfaces:
Serializable, Comparable<Class<? extends TraversalStrategy>>, TraversalStrategy<TraversalStrategy.OptimizationStrategy>, TraversalStrategy.OptimizationStrategy
public final class IdentityRemovalStrategy extends AbstractTraversalStrategy<TraversalStrategy.OptimizationStrategy> implements TraversalStrategy.OptimizationStrategy
IdentityRemovalStrategy looks for IdentityStep instances and removes them. If the identity step is labeled, its labels are added to the previous step. If the identity step is labeled, and it's the first step in the traversal, it stays.

Also for branch()/union() type steps an EndStep gets added which would lead to a traversal like: [UnionStep([[VertexStep(OUT,vertex), EndStep], [EndStep], [VertexStep(OUT,vertex), EndStep]])] if the identity() was removed. seems to make sense to account for that case so that the traversal gets to be: [UnionStep([[VertexStep(OUT,vertex), EndStep], [IdentityStep, EndStep], [VertexStep(OUT,vertex), EndStep]])] EndStep seems to just behave like an identity() in the above case, but perhaps it is more consistent to keep the identity() placeholder rather than a step that doesn't actually exist. Same applied to repeat() which would add RepeatEndStep, it's safe to keep RepeatStep([IdentityStep, RepeatEndStep] instead of leaving only RepeatEndStep.

Author:
Marko A. Rodriguez (http://markorodriguez.com)
See Also:
Example:
 __.out().identity().count()            // is replaced by __.out().count()
 __.in().identity().as("a")             // is replaced by __.in().as("a")
 __.identity().as("a").out()            // is replaced by __.as("a").out()