[prev in list] [next in list] [prev in thread] [next in thread]
List: log4j-dev
Subject: [jira] [Commented] (LOG4J2-761) Move performance tests from JUnit to log4j-perf
From: "Remko Popma (JIRA)" <jira () apache ! org>
Date: 2016-04-27 15:51:13
Message-ID: JIRA.12731633.1407105347000.48010.1461772273176 () Atlassian ! JIRA
[Download RAW message or body]
[ https://issues.apache.org/jira/browse/LOG4J2-761?page=com.atlassian.jira.plugin. \
system.issuetabpanels:comment-tabpanel&focusedCommentId=15260358#comment-15260358 ]
Remko Popma commented on LOG4J2-761:
------------------------------------
I don't remember exactly...
One drawback of moving these to log4j-perf is that they don't get run unless someone \
manually selects and runs them. We should look into [running some of the JMH \
benchmarks from a JUnit \
test|http://stackoverflow.com/questions/30485856/how-to-run-jmh-from-inside-junit-tests] \
and asserting some minimum performance level.
I often only run {{mvn clean package}} because the tests in the \
{{@Category(PerformanceTests.class)}} category are so slow... There are some valuable \
tests in there though: one of the deadlock tests caught a nasty bug I had in one of \
my attempts to make the OutputStreamAppender and subclasses garbage-free. That bug \
was not caught by the tests in {{mvn clean package}}...
I guess what I'm trying to say is that these tests need to be carefully checked one \
by one before we migrate them to the normal JUnit tests, converted to a \
(JUnit-driven?) JMH benchmark or leave them as is.
> Move performance tests from JUnit to log4j-perf
> -----------------------------------------------
>
> Key: LOG4J2-761
> URL: https://issues.apache.org/jira/browse/LOG4J2-761
> Project: Log4j 2
> Issue Type: Test
> Components: Core
> Affects Versions: 2.0.1
> Reporter: Matt Sicker
> Priority: Minor
> Labels: benchmark, test
>
> Firstly, all the JUnit tests marked with the category:
> {code}
> @Category(PerformanceTests.class)
> {code}
> should all be ported over to the log4j-perf module. The tests marked with this \
> category generally only output performance benchmark data and don't perform any \
> actual unit testing of their own. Now that we have several examples of how to use \
> [JMH|http://openjdk.java.net/projects/code-tools/jmh/], it would be a good idea to \
> clean up our unit tests as such. This way, we don't have to waste time in the build \
> performing benchmark tests that aren't even verified in any way.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: log4j-dev-unsubscribe@logging.apache.org
For additional commands, e-mail: log4j-dev-help@logging.apache.org
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic