[
https://issues.apache.org/jira/browse/MAPREDUCE-6368?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14547690#comment-14547690
]
Akira AJISAKA commented on MAPREDUCE-6368:
------------------------------------------
Thanks [~dhirajnilange] for reporting this issue. I'm thinking the condition
can be true.
{code}
float stepSize = samples.length / (float) numPartitions;
int last = -1;
for(int i = 1; i < numPartitions; ++i) {
int k = Math.round(stepSize * i);
while (last >= k && comparator.compare(samples[last], samples[k]) == 0) {
++k;
}
writer.append(samples[k], nullValue);
last = k;
}
{code}
{{k = Math.round(stepSize * i)}} can be equal to {{last = Math.round(stepSize *
(i-1))}} if {{stepSize}} is less than 1.
> Unreachable Java code
> ---------------------
>
> Key: MAPREDUCE-6368
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-6368
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2
> Affects Versions: 2.6.0
> Reporter: Dhiraj Nilange
> Priority: Minor
>
> Reference
> Class: org.apache.hadoop.mapreduce.lib.partition.InputSampler
> Method: writePartitionFile
> Line: 337
> The issue exists in the following code loop at line 337:-
> while (last >= k && comparator.compare(samples[last], samples[k]) == 0)
> {
> ++k;
> }
> The problem is that the first condition in the while loop (last >= k) will
> always be false. The value of 'last' will always be lesser than 'k' and hence
> the first condition will never evaluate to true. There is second condition as
> well but since it is appearing as AND condition, that will never be checked
> since the first condition itself is false. Hence this loop is not
> contributing towards the code output anyways. If this was intended to
> execute, then I guess it will need investigation. But from what I have
> noticed, it doesn't seem to harm the output of the method. In that case why
> even keep it there. We could very well remove it from the code. And if this
> was done with the some other intention, in that case this needs to be
> corrected as currently it is unreachable code. This issue very much exists in
> the release 2.6.0, I have not seen the release 2.7.0 source code, but it may
> very well exist in that as well (it's worth checking).
> Thanks & Regards,
> Dhiraj
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
