[
https://issues.apache.org/jira/browse/SPARK-4491?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14219181#comment-14219181
]
Sean Owen commented on SPARK-4491:
----------------------------------
If these are questions, best to close this and move to the mailing list?
I don't think you want to blindly take 'first' in case of a conflict. You want
to decide whether it's valid to skip them entirely (usually the case for
META-INF cruft) or merge the contents. Merge is a better default than either.
I don't think SBT is blaming anything on anyone, just pointing out that Maven
has different semantics. I think Maven's semantics are more practical but see
the point. You also see the solution there as well.
> Using sbt assembly with spark as dep requires Phd in sbt
> --------------------------------------------------------
>
> Key: SPARK-4491
> URL: https://issues.apache.org/jira/browse/SPARK-4491
> Project: Spark
> Issue Type: Question
> Reporter: sam
>
> I get the dreaded deduplicate error from sbt. I resolved the issue (I think,
> I managed to run the SimpleApp example) here
> http://stackoverflow.com/a/27018691/1586965
> My question is, is this wise? What is wrong with changing the `deduplicate`
> bit to `first`. Why isn't it this by default?
> If this isn't the way to make it work, please could someone provide an
> explanation of the correct way with .sbt examples. Having googled, every
> example I see is different because it changes depending on what deps the
> person has ... surely there has to be an automagic way of doing it (if my way
> isn't it)?
> One final point, SBT seems to be blaming Spark for causing the problem in
> their documentation: https://github.com/sbt/sbt-assembly is this fair? Is
> Spark doing something wrong in the way they build their jars? Or should SBT
> be renamed to CBT (Complicated Build Tool that will make you need Cognitive
> Behavioural Therapy after use).
> NOTE: Satire JFF, really I love both SBT & Spark :)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
