[
https://issues.apache.org/jira/browse/TINKERPOP-1862?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16383801#comment-16383801
]
ASF GitHub Bot commented on TINKERPOP-1862:
-------------------------------------------
GitHub user spmallette opened a pull request:
https://github.com/apache/tinkerpop/pull/804
TINKERPOP-1862 Messenger proper handling of Direction.BOTH
https://issues.apache.org/jira/browse/TINKERPOP-1862
This pulls in all changes from #801 and adds fixes to allow integration
tests in spark and giraph to pass.
This isn't completely ready for VOTE yet, but wanted to get it in place for
review by @PBGraff to make sure I didn't leave anything out by mistake. Also,
needed time to for docker tests to run. I will update this PR with a VOTE when
it's ready for general review.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/apache/tinkerpop TINKERPOP-1862
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/tinkerpop/pull/804.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 #804
----
commit f02ea338a8d039c50896f7c65ca57a695975fc43
Author: Graff, Philip B <philip.graff@...>
Date: 2018-01-04T18:27:44Z
TinkerMessenger proper handling of Direction.BOTH
commit 80c0b8465fe9bb6ddbe4522a36304c7ba054e909
Author: Graff, Philip B <philip.graff@...>
Date: 2018-01-14T02:32:29Z
Adding simple graph computer test of proper message passing in all 3
directions
commit 750677cf55d7445c3746870d86c6592f6cd3664d
Author: Graff, Philip B <philip.graff@...>
Date: 2018-01-14T04:49:27Z
Test cleanup - hopefully can run now
commit 153238b2a8a3fc5e051251662ef6ab80f72c659f
Author: Graff, Philip B <philip.graff@...>
Date: 2018-01-14T06:01:17Z
Additional check. Tests still fail - but I think this is a problem with the
implementations, not the test.
commit 5467a33f2e17ad22b736effebede124556f049ed
Author: Graff, Philip B <philip.graff@...>
Date: 2018-01-14T18:44:33Z
Test bug fix and refactoring repeated code
commit febe0ac8ecdf1828233017864b629a644c7260a8
Author: Graff, Philip B <philip.graff@...>
Date: 2018-02-25T17:22:39Z
Updating kitchen sink graph
commit f03c4df45473bafcf95d47f34467479f89af73a1
Author: Graff, Philip B <philip.graff@...>
Date: 2018-02-25T17:35:43Z
Updating message passing tests to use kitchen sink
commit b0b124b84c597079b271adb0dec2044f0c2ec9b3
Author: Graff, Philip B <philip.graff@...>
Date: 2018-02-25T21:15:40Z
Revert "Updating kitchen sink graph"
This reverts commit febe0ac8ecdf1828233017864b629a644c7260a8.
commit 0490a082d9b533cb7ef4a1a25bd25500844821a6
Author: Graff, Philip B <philip.graff@...>
Date: 2018-02-25T21:27:28Z
Proper update of kitchen sink through TinkerFactory
commit 027ae2771d3c67bb613c525a689fe10123686b45
Author: Graff, Philip B <philip.graff@...>
Date: 2018-02-28T16:38:33Z
Using "name" instead of "propertyin"
commit 26a5770efb288d60150cf9db60a5dd67568179f2
Author: Stephen Mallette <spmva@...>
Date: 2018-03-02T16:29:57Z
TINKERPOP-1862 Fix Messenger implementations for Spark/Giraph handling BOTH
These now behave like TinkerMessenger and in the case of BOTH pass the
message to the opposite vertex in the StarGraph
commit 7956b481b459a9998401abc8b02322f256a73ae7
Author: Stephen Mallette <spmva@...>
Date: 2018-03-02T16:33:21Z
TINKERPOP-1862 Updated changelog
----
> TinkerGraph VertexProgram message passing doesn't work properly when using
> Direction.BOTH
> -----------------------------------------------------------------------------------------
>
> Key: TINKERPOP-1862
> URL: https://issues.apache.org/jira/browse/TINKERPOP-1862
> Project: TinkerPop
> Issue Type: Bug
> Components: process, tinkergraph
> Affects Versions: 3.2.7
> Reporter: Philip Graff
> Priority: Major
>
> I think the messages are being sent properly in TinkerMessenger, but when I
> call messenger.receiveMessages(), the vertex is getting messages from the
> outVertex of their edges regardless of the edge direction. This is due to
> line 71 of TinkerMessenger (linked below) which calls
> Edge.vertices(direction).next(), thus getting the first result out of the
> Vertex iterator. For IN or OUT, this isn't a problem. But for BOTH, following
> line 124 of TinkerEdge (linked below), the outVertex is always returned
> first. TinkerMessenger needs to be modified to return the correct vertex (I
> think it's the one that != this.vertex).
> https://github.com/apache/tinkerpop/blob/master/tinkergraph-gremlin/src/main/java/org/apache/tinkerpop/gremlin/tinkergraph/process/computer/TinkerMessenger.java#L71
> https://github.com/apache/tinkerpop/blob/master/tinkergraph-gremlin/src/main/java/org/apache/tinkerpop/gremlin/tinkergraph/structure/TinkerEdge.java#L124
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
