[SPARK-5600] [core] Clean up FsHistoryProvider test, fix app sort order.

[vanzin]

Clean up some test setup code to remove duplicate instantiation of the
provider. Also make sure unfinished apps are sorted correctly.

[SparkQA]

Test build #26768 has started for PR 4370 at commit 298371c.

• This patch merges cleanly.

[SparkQA]

Test build #26768 has finished for PR 4370 at commit 298371c.

• This patch fails PySpark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/26768/
Test FAILed.

[vanzin]

Jenkins, retest this please.

[SparkQA]

Test build #26779 has started for PR 4370 at commit 298371c.

• This patch merges cleanly.

[SparkQA]

Test build #26779 has finished for PR 4370 at commit 298371c.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/26779/
Test PASSed.

[srowen]

The rest LGTM, but, isn't this defining a different and inconsistent sort order? For example if the first argument's start and end time are after and before the other's start and times, respectively, then the method returns false both ways, when one has to be true. The existing logic is to sort by endTime descending, then startTime descending, so I think it's more like...

if (i1.endTime != i2.endTime) {
  i1.endTime > i2.endTime
} else {
  i1.startTime > i2.startTime
}

[SparkQA]

Test build #26939 has started for PR 4370 at commit 0d048d5.

• This patch merges cleanly.

[SparkQA]

Test build #26939 has finished for PR 4370 at commit 0d048d5.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds the following public classes (experimental):
  • "public class " + className + extendsText + " implements java.io.Serializable
  • case class RegisterExecutor(

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/26939/
Test PASSed.

[andrewor14]

LGTM I'm merging this into master 1.3 thanks @vanzin

[srowen]

I don't imagine it probably matters, but this should be > instead of >= I think? or else this says that two equal intervals should both precede the other. I bet it doesn't make the sort wrong, just might affect whether the sort is stable, but who cares here.