[
https://issues.apache.org/jira/browse/HADOOP-10865?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15140314#comment-15140314
]
Masatake Iwasaki commented on HADOOP-10865:
-------------------------------------------
The patch still applies and I agree to add the benchmark. Just a few comments.
{noformat}
Running org.apache.hadoop.util.TestDataChecksum
Tests run: 5, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 10.174 sec <<<
FAILURE! - in org.apache.hadoop.util.TestDataChecksum
testCrc32(org.apache.hadoop.util.TestDataChecksum) Time elapsed: 0.11 sec <<<
FAILURE!
java.lang.AssertionError: NativeCrc32 is not available
at
org.apache.hadoop.util.Crc32PerformanceTest.<init>(Crc32PerformanceTest.java:122)
at
org.apache.hadoop.util.TestDataChecksum.testCrc32(TestDataChecksum.java:203)
{noformat}
The {{TestDataChecksum#testCrc32}} case should be skipped in non-native profile
by calling {{GenericTestUtils#assumeInNativeProfile}}.
{code}
public void testCrc32() throws Exception {
new Crc32PerformanceTest(8, 3, true).run();
new Crc32PerformanceTest(8, 3, false).run();
}
{code}
Do we need the both? Running only {{Crc32PerformanceTest(8, 3, true)}} seems to
be ok. It makes test time shorter.
> Add a Crc32 chunked verification benchmark for both directly and non-directly
> buffer cases
> ------------------------------------------------------------------------------------------
>
> Key: HADOOP-10865
> URL: https://issues.apache.org/jira/browse/HADOOP-10865
> Project: Hadoop Common
> Issue Type: Improvement
> Components: util
> Reporter: Tsz Wo Nicholas Sze
> Assignee: Tsz Wo Nicholas Sze
> Priority: Minor
> Labels: BB2015-05-TBR
> Attachments: HADOOP-10865.002.patch, c10865_20140717.patch
>
>
> Currently, it is not easy to compare Crc32 chunked verification
> implementations. Let's add a benchmark.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
