[
https://issues.apache.org/jira/browse/FLINK-1974?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14531287#comment-14531287
]
ASF GitHub Bot commented on FLINK-1974:
---------------------------------------
Github user jkirsch commented on a diff in the pull request:
https://github.com/apache/flink/pull/652#discussion_r29795245
--- Diff:
flink-core/src/main/java/org/apache/flink/api/common/JobExecutionResult.java ---
@@ -47,12 +48,23 @@ public JobExecutionResult(JobID jobID, long netRuntime,
Map<String, Object> accu
* Gets the net execution time of the job, i.e., the execution time in
the parallel system,
* without the pre-flight steps like the optimizer.
*
- * @return The net execution time.
+ * @return The net execution time in milliseconds.
*/
public long getNetRuntime() {
return this.netRuntime;
}
+ /**
+ * Gets the net execution time of the job, i.e., the execution time in
the parallel system,
+ * without the pre-flight steps like the optimizer in a desired unit.
+ *
+ * @param desiredUnit the unit of the <tt>NetRuntime</tt>
+ * @return The net execution time in the desired unit.
+ */
+ public long elapsedNetRuntime(TimeUnit desiredUnit) {
--- End diff --
Sure, I just based it on the [Guava StopWatch
](http://docs.guava-libraries.googlecode.com/git/javadoc/com/google/common/base/Stopwatch.html)
functionality which has elapsed.
In hindsight I noticed that the runtime for the
org.apache.flink.runtime.jobmanager is now reporting the runtime in seconds
`org.apache.flink.runtime.jobmanager.JobInfo` which is a change from the 0.8.1
release, which I used and noticed the missing documentation.
```scala
def duration: Long = {
if(end != -1){
(end - start)/1000
}else{
-1
}
```
introduced with commit 241c1ca4db1e5f4b3f81b3cdf694246062f032d1
So I either update the code to the current state (seconds) or propose for
high accuracy to revert back to the old behaviour ?
> JobExecutionResult NetRuntime - document result type
> ----------------------------------------------------
>
> Key: FLINK-1974
> URL: https://issues.apache.org/jira/browse/FLINK-1974
> Project: Flink
> Issue Type: Improvement
> Components: Documentation
> Affects Versions: 0.9
> Environment: The JobExecutionResult stores the net execution times in
> milliseconds, but does not say so.
> A simple fix is to add it to the JavaDoc .. a more complete fix would be to
> change the method name to getNetRuntimeMs or even use an auto conversion,
> such as getElapsedNetRuntime(Timeunit) which would autoconvert
> Reporter: Johannes
> Assignee: Johannes
> Priority: Trivial
> Labels: easyfix
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
