[
https://issues.apache.org/jira/browse/HCATALOG-569?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13531692#comment-13531692
]
Travis Crawford commented on HCATALOG-569:
------------------------------------------
Re (1) - I'm not fully up-to-date on changes with MR2. With the YARN changes,
I'm not sure if there's a drop-in replacement. There is {{hadoop-minicluster}}
which you can find on maven central - will that work for us?
Re (2) - there are three tests actually that will fail due to the circular
dependency:
{code}
fynn-2:hcatalog-ro travis$ grep -l -r HCatLoader core/src/test
core/src/test/java/org/apache/hcatalog/mapreduce/TestHCatHiveCompatibility.java
core/src/test/java/org/apache/hcatalog/mapreduce/TestHCatHiveThriftCompatibility.java
core/src/test/java/org/apache/hcatalog/mapreduce/TestSequenceFileReadWrite.java
fynn-2:hcatalog-ro travis$
{code}
{{TestHCatHiveCompatibility}} fails with a compile error as it uses
{{HCatLoader.class.getName()}} in a pig load statement. The other usages
reference HCatLoader in pig statement strings and will fail at runtime.
I *think* the best approach here is updating these tests to use input/output
formats, instead of pig, as the goal of core tests are to test the MR-native
functionality. Of course, we could jump through some hoops and publish
{{hcatalog-pig-adapter-testing.jar}} that core could use during tests, but as
you point out that mixes the layers in a way that's not ideal.
Re: (3) the number I'm giving to your core test classes item, yes, we do want
subprojects to use {{HCatBaseTest}} and friends. Suggestions on how best to do
this? Some projects publish a "testing" artifact and perhaps we could follow
that pattern.
> Add support for building HCatalog with Maven
> --------------------------------------------
>
> Key: HCATALOG-569
> URL: https://issues.apache.org/jira/browse/HCATALOG-569
> Project: HCatalog
> Issue Type: Bug
> Reporter: Arup Malakar
>
> Hi,
> I see that:
> {code}
> mvn clean package -Phadoop23
> {code}
> fails in trunk.
> Couple of observations:
> 1. *Missing hadoop-test for hadoop23 profile*:
> Test code in core depends on _MiniMRCluster_, and the compilation fails when
> mvn is ran with -Phadoop23 profile.
> I see that _MiniMRCluster_ is present in _hadoop-test_ artifact. But I don't
> see a hadoop-test artifact being available for hadoop 23:
> https://repository.apache.org/index.html#nexus-search;quick~hadoop-test
> Question: Is it okay to include _1.0.3_ version of hadoop-test in case of
> hadoop23 profile as well?
> 2. *Circular dependency between core and hcatalog-pig-adapter*
> I observed core has test cases which requires *HCatLoader* which is available
> in *hcatalog-pig-adapter*. On the other hand *hcatalog-pig-adapter* depends
> on core.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
