SOLR-16040: Fix split packages in hdfs module

[risdenk]

https://issues.apache.org/jira/browse/SOLR-16040

Its broken down into multiple commits just as I worked through it. The only weird commit is ce7d256c89eca082a3ebe59595a9d61e098871bc where there were a bunch of changes needed to open up update and transaction classes to allow the new hdfs package to work. Everything before was package private so it worked in the same package.

Tested with:

./gradlew check
./gradlew test --tests "*hdfs*" --tests "*Hdfs*" --tests "*HDFS*" --tests "*Hadoop*" -Ptests.jvms=4 -Ptests.jvmargs=-XX:TieredStopAtLevel=1 -Ptests.nightly=true -Ptests.awaitsfix=false -Ptests.badapples=false -Ptests.file.encoding=UTF-8

[dsmiley]

While you're at it, could you please rename pertinent tests that aren't using camel case for Hdfs like HDFSCollectionsAPITest ? Maybe non-test classes too. This is the standard I've seen most projects gravitate towards instead of it being haphazard.

[janhoy]

I love to see how eating our own dogfood improves our base classes by detecting private access for crucial methods and variables, that would prevent users to create their own custom plugins for the same.

Did a quick browse only.

[risdenk]

@dsmiley I think I fixed all the class names in 1f47ffe

[risdenk]

@dsmiley fixed Filesystem in f5249ad

[HoustonPutman]

This looks good to me with a changelog entry. Though I can't speak for the SolrResourceLoader class.

[HoustonPutman]

Did anything else get overridden in the force push?

[risdenk]

@HoustonPutman the force push wasn't the issue. I didn't pull the branch before making local changes to add the solr/CHANGES.txt. The only change made through the UI was your documentation change.

[HoustonPutman]

Ahh ok, thanks for explaining 🙂