jmeter-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Felix Schumacher <felix.schumac...@internetallee.de>
Subject Re: JMeter code coverage report: should it include *test* classes?
Date Sat, 18 May 2019 10:41:56 GMT

Am 15.05.19 um 19:54 schrieb Vladimir Sitnikov:
> Hi,
>
> I as a part of implementing Gradle build script I identified an issue with
> the current Ant build.
>
> Current script calculates "coverage" and "complexity" for TEST source code.
> For instance:
> https://codecov.io/gh/apache/jmeter/tree/c63d8c15c0e1c1df4799af8941fd488c19cdcf98/test
>
> It adds 15'145 "lines hit", 294 "lines partial", however it is apparently
> is a garbage data. Does it really make sense to collect coverage for test
> classes?

I don't think that we need test coverage for the test classes. It was
probably added by mistake.

Felix

>
> Current Codecov report (see
> https://github.com/apache/jmeter/pull/448#issuecomment-470432467 ) for
> "Gradle migration" shows "9% decrease" coverage, however I believe the key
> difference is Gradle/JaCoCo do not count test classes as "lines hit" by
> default.
>
> So I assume Gradle-driven code coverage is good enough.
>
> Vladimir
>

Mime
View raw message