[
https://issues.apache.org/jira/browse/MAPREDUCE-7197?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16899900#comment-16899900
]
Szilard Nemeth commented on MAPREDUCE-7197:
-------------------------------------------
Hi [~adam.antal]!
Thanks for this patch, it's quite extensive and precise.
However, I have some comments:
1. In
org.apache.hadoop.mapreduce.jobhistory.TestJobHistoryEventHandler#testAMStartedEvent:
There are 2 code blocks you modified:
a.
{code:java}
assertThat(100).isEqualTo(mi.getJobIndexInfo().getSubmitTime());
assertThat(200).isEqualTo(mi.getJobIndexInfo().getJobStartTime());
assertThat(100).isEqualTo(mi.getJobSummary().getJobSubmitTime());
assertThat(200).isEqualTo(mi.getJobSummary().getJobLaunchTime());
{code}
b.
{code:java}
assertThat(100).isEqualTo(mi.getJobIndexInfo().getSubmitTime());
assertThat(200).isEqualTo(mi.getJobIndexInfo().getJobStartTime());
assertThat(100).isEqualTo(mi.getJobSummary().getJobSubmitTime());
assertThat(200).isEqualTo(mi.getJobSummary().getJobLaunchTime());
{code}
Even though the original code had the expected and actual values interchanged,
you should change the expected values with actual for these, for example:
{code:java}
assertThat(mi.getJobIndexInfo().getSubmitTime()).isEqualTo(100); --> Expected
value was 100.
{code}
2. In
org.apache.hadoop.mapreduce.v2.app.TestRuntimeEstimators#coreTestEstimator:
You removed one meaningful assertion with your patch:
You have:
{code:java}
assertThat(speculator.getProportionRunningTasksSpeculatable())
.isCloseTo(0.1, offset(0.00001));
{code}
And we had:
{code:java}
Assert.assertEquals(speculator.getProportionRunningTasksSpeculatable(),
0.1, 0.00001);
Assert.assertEquals(speculator.getProportionTotalTasksSpeculatable(),
0.001, 0.00001);
{code}
, so the assertion for speculator.getProportionTotalTasksSpeculatable() is
missing.
3. In
org.apache.hadoop.mapreduce.v2.app.job.impl.TestTaskAttempt#testContainerCleanedWhileRunning:
You have this change:
{code:java}
assertThat(taImpl.getState())
.withFailMessage("Task attempt is not in running state")
.isEqualTo(TaskAttemptState.RUNNING);
{code}
I'd prefer to move the call to isEqualTo above withFailMessage.
You have similar call structures throughout this class, please also change all
of them!
4. In
org.apache.hadoop.mapreduce.v2.app.job.impl.TestTaskAttempt#testFetchFailureAttemptFinishTime:
This assertion is weird, the message is not in line with the state type, please
fix it! :
{code:java}
assertThat(taImpl.getState())
.withFailMessage("Task attempt is not in Too Many Fetch Failure state")
.isEqualTo(TaskAttemptState.FAILED);
{code}
5. In
org.apache.hadoop.mapreduce.v2.app.job.impl.TestTaskAttempt#testContainerKillAfterAssigned:
There's a typo (not introduced by you) in this assertion message:
{code:java}
assertThat(taImpl.getInternalState())
.withFailMessage("Task attempt is not in assinged state")
.isEqualTo(TaskAttemptStateInternal.ASSIGNED);
{code}
6. In
org.apache.hadoop.mapreduce.v2.app.job.impl.TestTaskAttempt#testTaskAttemptDiagnosticEventOnFinishing:
I like that you fixed some assertion message here:
{code:java}
assertThat(taImpl.getState())
.withFailMessage("Task attempt is not in succeeded state")
.isEqualTo(TaskAttemptState.SUCCEEDED);
{code}
However, please use the name of the state with uppercase letters. You have one
more occurrence of this in the same method.
7. In org.apache.hadoop.mapred.TestJobConf:
I can see some updates regarding enum fields (asserting them directly without
using .name() ) but I can't see assertEquals rewritten to assertThat for a
bunch of occurrences in the class, please fix this!
8. In org.apache.hadoop.mapreduce.task.reduce.TestMergeManager:
Your 2 changes in this class are using assertNull. Can you please use
assertThat + isNull instead?
You have an occurrence of this in the following palces as well:
{code:java}
- org.apache.hadoop.mapred.TestTaskLog#testTaskLogWithoutTaskLogDir
- In class org.apache.hadoop.mapreduce.security.TestTokenCache
- org.apache.hadoop.mapred.uploader.TestFrameworkUploader#testHelp
{code}
9. In
org.apache.hadoop.mapreduce.v2.hs.TestJHSDelegationTokenSecretManager#stopAndCleanSecretManager:
You have this assertion:
{code:java}
assertThat(mgr.getAllTokens().size())
.withFailMessage("Secret manager should not contain keys").isZero();
{code}
The original assertion was:
{code:java}
Assert.assertEquals("Secret manager should not contain tokens",
mgr.getAllTokens().size(), 0);
{code}
The assertion message should have the word "tokens" instead of "keys", I think.
10. In
org.apache.hadoop.mapreduce.v2.hs.TestJobHistoryEntities#testCompletedJob:
You rewritten one assertEquals to assertTrue. Can you use assertThat instead?
11. In org.apache.hadoop.mapred.TestMultiFileSplit#testReadWrite:
You started to use assertArrayEquals. Can you verify if assertJ has any method
to test arrays / lists?
12. In org.apache.hadoop.mapreduce.v2.TestUberAM#testFailingMapper:
You have one change that introduced Assert.assertFalse. Please use some assertJ
method for this instead!
You have a similar ones in the following places (assertTrue / assertFalse):
{code:java}
-
org.apache.hadoop.mapred.nativetask.combinertest.CombinerTest#testWordCountCombiner
- org.apache.hadoop.mapred.nativetask.compresstest.CompressTest (throughout the
class)
- org.apache.hadoop.mapred.nativetask.kvtest.KVTest#testKVCompability
-
org.apache.hadoop.mapred.nativetask.combinertest.LargeKVCombinerTest#testLargeValueCombiner
- org.apache.hadoop.mapred.nativetask.nonsorttest.NonSortTest#nonSortTest
-
org.apache.hadoop.mapred.nativetask.combinertest.OldAPICombinerTest#testWordCountCombinerWithOldAPI
- TestByteBufferReadWrite (throughout the class)
{code}
That's all I have!
Please fix these and upload a new patch. Obviously, I will only check the diff
between the 2 patches :D
> Fix order of actual and expected expression in assert statements
> ----------------------------------------------------------------
>
> Key: MAPREDUCE-7197
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-7197
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Affects Versions: 3.2.0
> Reporter: Adam Antal
> Assignee: Adam Antal
> Priority: Major
> Attachments: MAPREDUCE-7197.001.patch, MAPREDUCE-7197.002.patch,
> MAPREDUCE-7197.003.patch, MAPREDUCE-7197.004.patch, MAPREDUCE-7197.004.patch,
> old-approx-from-HADOOP-12693
>
>
> Fix order of actual and expected expression in assert statements which gives
> misleading message when test case fails. Attached file has some of the places
> where it is placed wrongly.
> {code:java}
> [ERROR]
> testNodeRemovalGracefully(org.apache.hadoop.yarn.server.resourcemanager.TestResourceTrackerService)
> Time elapsed: 3.385 s <<< FAILURE!
> java.lang.AssertionError: Shutdown nodes should be 0 now expected:<1> but
> was:<0>
> {code}
> For long term, [AssertJ|http://joel-costigliola.github.io/assertj/] can be
> used for new test cases which avoids such mistakes.
> This is a follow-up Jira on the fix for the MR project.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
