[
https://issues.apache.org/jira/browse/TINKERPOP-1502?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15667899#comment-15667899
]
ASF GitHub Bot commented on TINKERPOP-1502:
-------------------------------------------
Github user dkuppitz commented on the issue:
https://github.com/apache/tinkerpop/pull/495
Since this PR touches `FilterRankingStrategyTest`s, it would be nice to
have this issue fixed as well (and have some tests added to cover this
scenario):
```
gremlin> g = TinkerFactory.createModern().traversal()
==>graphtraversalsource[tinkergraph[vertices:6 edges:6], standard]
gremlin> g.V(1).as("a").both().both().has("age").where(neq("a")).by("age")
The property does not exist as the key has no associated value for the
provided element: v[5]:age
Type ':help' or ':h' for help.
Display stack trace? [yN]N
gremlin>
g.V(1).as("a").both().both().has("age").where(neq("a")).by("age").explain()
==>Traversal Explanation
===========================================================================================================================================================================================================================================
Original Traversal [GraphStep(vertex,[1])@[a],
VertexStep(BOTH,vertex), VertexStep(BOTH,vertex),
TraversalFilterStep([PropertiesStep([age],value)]),
WherePredicateStep(neq(a),[value(age)])]
...
FilterRankingStrategy [O] [GraphStep(vertex,[1])@[a],
VertexStep(BOTH,vertex), VertexStep(BOTH,vertex),
WherePredicateStep(neq(a),[value(age)]),
TraversalFilterStep([PropertiesStep([age],value)])]
...
Final Traversal [TinkerGraphStep(vertex,[1])@[a],
VertexStep(BOTH,vertex), VertexStep(BOTH,vertex),
WherePredicateStep(neq(a),[value(age)]), NoOpBarrierStep(2500),
TraversalFilterStep([PropertiesStep([age],property)])]
```
> Chained has()-steps should simply left-append HasContainers in Gremlin-Java.
> ----------------------------------------------------------------------------
>
> Key: TINKERPOP-1502
> URL: https://issues.apache.org/jira/browse/TINKERPOP-1502
> Project: TinkerPop
> Issue Type: Improvement
> Components: process
> Affects Versions: 3.2.2
> Reporter: Marko A. Rodriguez
>
> In Gremlin-Java, {{g.V().has(a).has(b).has(c).out()}} is originally
> represented as {{[GraphStep,HasStep(a),HasStep(b),HasStep(c),VertexStep]}}.
> Ultimately, {{InlineFilterStrategy}} or most provider strategies will turn
> such {{HasStep}}-chains into {{[GraphStep,HasStep(a,b,c),VertexStep]}}. That
> is, strategies fold {{has()}}-steps "left" and delete "right" {{has()}}-steps
> and left propagates their labels (i.e. clock cycles). I think that
> {{GraphTraversal}} should simply do this:
> {code}
> public GraphTraversal has(whateves) {
> if(this.getEndStep() instanceof HasStep)
> this.getEndSte().addHasContainer(new HasContainer(whateves))
> else
> this.addStep(new HasStep(new HasContainer(whateves)));
> this.bytecode.addStep("has",whateves);
> return this;
> }
> {code}
> In essence, a "write time" optimization can be done. Given that chains of
> {{has()}}'s is super common, this can save significant clock-cycles in the
> long run of a production application.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
