[
https://issues.apache.org/jira/browse/SOLR-11749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16375717#comment-16375717
]
Jason Gerlowski commented on SOLR-11749:
----------------------------------------
I've attached an update patch with finished test suites for {{bin/solr
create_collection}} and {{bin/solr delete}} (collection).
*PROS*
- Following in xUnit footsteps makes the tests readable and understandable.
- Adding individual test cases is pretty easy.
*CONS*
- no parallelism makes test running slow. The latest patch runs about 26 tests
in about 3 and a half minutes locally for me. If there's a clean way to do
parallelism in bash, it'd be new to me. I'm open to ideas. This is concerning,
but not a showstopper IMO. Slow {{bin/solr}} tests are better than none at all.
We probably shouldn't hook them into the normal {{ant-test}} target until we
find some way to improve the speed though.
- Many of the more useful {{@Before}}/{{@After}} actions end up talking to ZK
to list all the collections/configs, etc. This really adds to the test-suite
runtime. Probably a good place for optimization if we need to cut down the
runtime.
- Implementing all of my bash logic in Windows/batch is a little scary. Both
because I don't know batch as well as bash, but also because I only have
intermittent access to a Windows OS.
*NEXT STEPS*
- Add support for single-test/single-suite running.
- Add Windows/batch equivalents
- Expand test coverage.
*QUESTIONS*
- The tests I've been able to write so far assert on a lot of things. Some
assert on the actual text output of the script. Some assert on changes to
global state in ZK (created collections/configs, etc.). I think a mix of both
is healthy. But if you had tests of a certain type in mind when you created
this JIRA [~cpoerschke], let me know and I can make sure I get coverage for the
things you're interested in.
Ideally, I would like to merge this once I get a Windows equivalent of the work
I've done so far. The current test coverage is limited, but getting even
limited coverage merged is beneficial, as some tests are better than none. It
also allow would allow other interested devs to contribute tests in parallel.
So getting to parity on Windows parity is my top priority.
Also, gonna assign this JIRA to myself, unless others want it.
> regression-test-like functionality for bin/solr*
> ------------------------------------------------
>
> Key: SOLR-11749
> URL: https://issues.apache.org/jira/browse/SOLR-11749
> Project: Solr
> Issue Type: Wish
> Security Level: Public(Default Security Level. Issues are Public)
> Components: scripts and tools
> Reporter: Christine Poerschke
> Priority: Minor
> Attachments: SOLR-11749.patch, SOLR-11749.patch, SOLR-11749.patch,
> test-output.txt
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
