[SPARK-20708][CORE] Make addExclusionRules up-to-date#17947
[SPARK-20708][CORE] Make addExclusionRules up-to-date#17947dongjoon-hyun wants to merge 5 commits intoapache:masterfrom dongjoon-hyun:SPARK-20708
addExclusionRules up-to-date#17947Conversation
|
Test build #76793 has finished for PR 17947 at commit
|
|
Hi, @srowen . |
|
Retest this please. |
|
Test build #77098 has finished for PR 17947 at commit
|
|
Retest this please. |
|
Test build #77103 has finished for PR 17947 at commit
|
|
cc @brkyvz to verify the change. |
|
Thank you, @jiangxb1987 . |
|
Could you review this please, @brkyvz ? |
|
|
||
| val coordinates = | ||
| components.map(comp => s"org.apache.spark:spark-${comp}2.10:1.2.0").mkString(",") + | ||
| components.map(comp => s"org.apache.spark:spark-${comp}2.10:2.1.1").mkString(",") + |
There was a problem hiding this comment.
maybe also change Scala version too, since we will remove 2.10 soon
There was a problem hiding this comment.
Sure! Thank you for review.
| test("neglects Spark and Spark's dependencies") { | ||
| val components = Seq("catalyst_", "core_", "graphx_", "hive_", "mllib_", "repl_", | ||
| "sql_", "streaming_", "yarn_", "network-common_", "network-shuffle_", "network-yarn_") | ||
| val components = Seq("catalyst_", "core_", "graphx_", "hive_", "hive-thriftserver_", |
There was a problem hiding this comment.
since the hive, mesos stuff is not in the assembly jar by default, I don't think we need to exclude those
There was a problem hiding this comment.
I see. I'll remove hive, hive-thriftserver, mesos, and yarn.
There was a problem hiding this comment.
Maybe, network-yarn_, too.
|
We shouldn't neglect dependencies that are not in the assembly jar by default. Otherwise thanks for doing this! |
| // spark-streaming_2.1x and spark-streaming-kafka-0-8-assembly_2.1x | ||
| val components = Seq("catalyst_", "core_", "graphx_", "hive_", "mllib_", "repl_", | ||
| "sql_", "streaming_", "yarn_", "network-common_", "network-shuffle_", "network-yarn_") | ||
| val components = Seq("catalyst_", "core_", "graphx_", "launcher_", "mllib_", "mllib-local_", |
There was a problem hiding this comment.
can we also make this a static variable and re-use in tests?
There was a problem hiding this comment.
otherwise it's hard to keep both in sync
|
|
||
| val coordinates = | ||
| components.map(comp => s"org.apache.spark:spark-${comp}2.10:1.2.0").mkString(",") + | ||
| components.map(comp => s"org.apache.spark:spark-${comp}2.11:2.1.1").mkString(",") + |
|
Test build #77554 has finished for PR 17947 at commit
|
|
Test build #77560 has finished for PR 17947 at commit
|
|
Hi, @brkyvz . |
| // We need to specify each component explicitly, otherwise we miss spark-streaming-kafka-0-8 and | ||
| // other spark-streaming utility components. Underscore is there to differentiate between | ||
| // spark-streaming_2.1x and spark-streaming-kafka-0-8-assembly_2.1x | ||
| val components = Seq("catalyst_", "core_", "graphx_", "launcher_", "mllib_", "mllib-local_", |
There was a problem hiding this comment.
since it's in global scope now can we rename it to something more meaningful?
IVY_DEFAULT_EXCLUDES
Also add to the docs, since it has lost its context from the method and the docs now
|
@brkyvz . Sure. I renamed it into |
|
Test build #77614 has finished for PR 17947 at commit
|
|
Test build #77615 has finished for PR 17947 at commit
|
|
Thanks @dongjoon-hyun LGTM. Merging to master |
|
Thank you so much, @brkyvz . |
What changes were proposed in this pull request?
Since SPARK-9263,
resolveMavenCoordinatesignores Spark and Spark's dependencies by usingaddExclusionRules. This PR aims to make addExclusionRules up-to-date to neglect correctly because it fails to neglect some components like the following.mllib (correct)
mllib-local (wrong)
How was this patch tested?
Pass the Jenkins with a updated test case.