[
https://issues.apache.org/jira/browse/TINKERPOP-1627?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15868689#comment-15868689
]
ASF GitHub Bot commented on TINKERPOP-1627:
-------------------------------------------
GitHub user okram opened a pull request:
https://github.com/apache/tinkerpop/pull/559
TINKERPOP-1627: LazyBarrierStrategy should not append an end barrier.
https://issues.apache.org/jira/browse/TINKERPOP-1627
You don't want `LazyBarrierStrategy` appending `NoOpBarriers` to the root
traversal. It was doing that because I was trying to be clever and have it
consider `RemoteTraversal` I/O (via bulking). Dumb. `RemoteStrategy` should be
responsible for that concern. Fixed.
VOTE +1.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/apache/tinkerpop TINKERPOP-1627
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/tinkerpop/pull/559.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #559
----
commit 080306b099b3ca5dbf5bf5041447fd1a0557b9cb
Author: Marko A. Rodriguez <[email protected]>
Date: 2017-02-15T22:07:53Z
Fixed an optimization bug in LazyBarrierStrategy. RemoteStrategy should be
responsible for network I/O concerns (and using bulking as the optimization).
----
> LazyBarrierStrategy should not append an end barrier.
> -----------------------------------------------------
>
> Key: TINKERPOP-1627
> URL: https://issues.apache.org/jira/browse/TINKERPOP-1627
> Project: TinkerPop
> Issue Type: Improvement
> Components: process
> Reporter: Marko A. Rodriguez
> Assignee: Marko A. Rodriguez
> Fix For: 3.2.5
>
>
> {code}
> gremlin> g.V().out().explain()
> ==>Traversal Explanation
> ==============================================================================================================
> Original Traversal [GraphStep(vertex,[]),
> VertexStep(OUT,vertex)]
> ConnectiveStrategy [D] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex)]
> RangeByIsCountStrategy [O] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex)]
> RepeatUnrollStrategy [O] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex)]
> MatchPredicateStrategy [O] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex)]
> PathRetractionStrategy [O] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex)]
> IncidentToAdjacentStrategy [O] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex)]
> FilterRankingStrategy [O] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex)]
> InlineFilterStrategy [O] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex)]
> AdjacentToIncidentStrategy [O] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex)]
> LazyBarrierStrategy [O] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex), NoOpBarrierStep(2500)]
> TinkerGraphCountStrategy [P] [GraphStep(vertex,[]),
> VertexStep(OUT,vertex), NoOpBarrierStep(2500)]
> TinkerGraphStepStrategy [P] [TinkerGraphStep(vertex,[]),
> VertexStep(OUT,vertex), NoOpBarrierStep(2500)]
> ProfileStrategy [F] [TinkerGraphStep(vertex,[]),
> VertexStep(OUT,vertex), NoOpBarrierStep(2500)]
> StandardVerificationStrategy [V] [TinkerGraphStep(vertex,[]),
> VertexStep(OUT,vertex), NoOpBarrierStep(2500)]
> Final Traversal [TinkerGraphStep(vertex,[]),
> VertexStep(OUT,vertex), NoOpBarrierStep(2500)]
> {code}
> The only time this is cool is for {{RemoteTraversal}} because you want to
> barrier as much as you can to reduce over the wire costs. However,
> {{RemoteStrategy}} should append an end-barrier. No other strategy should do
> this --- its just wasted clock cycles.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
