[
https://issues.apache.org/jira/browse/CALCITE-1561?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15903774#comment-15903774
]
Eli Levine commented on CALCITE-1561:
-------------------------------------
[~julianhyde], this is a strange beast. I can't repro on either Mac or Linux.
However, I remember that I was able to repro the inconsistent failures before,
so not sure what's going on here. This test uses the PigTest class to execute
Pig in local mode and perform data validation. PigTest creates data files in
/tmp for result validation. I suspect these issues are related to local FS
reads/writes but I can't confirm consistently, since I can't repro.
What I remember from the last time I saw these failures is that one would only
see the once. Everything worked fine on subsequent runs. What I will do is
introduce a dummy run of PigTest before real tests are run in
PigRelBuilderStyleTest. Let's see if it cuts down the noise. Will send a PR
shortly.
> Intermittent test failures
> --------------------------
>
> Key: CALCITE-1561
> URL: https://issues.apache.org/jira/browse/CALCITE-1561
> Project: Calcite
> Issue Type: Bug
> Reporter: Julian Hyde
> Assignee: Julian Hyde
>
> A few tests are currently failing intermittently.
> 1. UdfTest
> {noformat}
> UdfTest.testUserDefinedFunction:162
> Expected: is <9>
> but: was <10>
> {noformat}
> 2. StreamTest.testInfiniteStreamsDoNotBufferInMemory
> {noformat}
> FAILURE! - in org.apache.calcite.test.StreamTest
> testInfiniteStreamsDoNotBufferInMemory(org.apache.calcite.test.StreamTest)
> Time elapsed: 0.218 sec <<< ERROR!
> java.lang.RuntimeException: exception while executing [select stream * from
> orders]
> at
> org.apache.calcite.test.StreamTest.testInfiniteStreamsDoNotBufferInMemory(StreamTest.java:237)
> Caused by: java.util.NoSuchElementException
> at
> org.apache.calcite.test.StreamTest.testInfiniteStreamsDoNotBufferInMemory(StreamTest.java:237)
> {noformat}
> I don't consider any of them serious -- they are test issues, not product
> correctness issues -- but they are inconvenient because false negatives waste
> developer time.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
