Re: [logging-log4j2] 02/02: Use final and simpler hashcode generation through Objects.

2022-01-12 Thread Carter Kozak
I'd prefer if we didn't incur implicit array allocation cost generating a hash code. My preference is to keep the original implementation. -ck On Wed, Jan 12, 2022, at 08:50, ggreg...@apache.org wrote: > This is an automated email from the ASF dual-hosted git repository. > > ggregory pushed a c

Re: [logging-log4j2] 02/04: Our convention is to make test classes public.

2022-01-12 Thread Matt Sicker
I'll note that the convention from JUnit 4 is to make them public; JUnit 5 encourages package-private tests instead for some reason, and that's the default template for JUnit 5 tests in IntelliJ. I do like consistency, though! On Wed, Jan 12, 2022 at 11:01 AM wrote: > > This is an automated email

Re: [logging-log4j2] 02/04: Our convention is to make test classes public.

2022-01-12 Thread Gary Gregory
Hi Matt, Porting to Junit 5 would be nice. I'm not sure how to deal with all our Junit 4 rules though. Gary On Wed, Jan 12, 2022, 13:07 Matt Sicker wrote: > I'll note that the convention from JUnit 4 is to make them public; > JUnit 5 encourages package-private tests instead for some reason, an

Re: [logging-log4j2] 02/04: Our convention is to make test classes public.

2022-01-12 Thread Carter Kozak
+1 I prefer minimum visibility by default for the same reason I prefer to make everything final by default: It gives us more freedom to change later on. This doesn't directly apply to tests, but it's nice when a convention applies globally. Most projects don't make junit5 tests public, so ther

Re: [CI][UNSTABLE] Logging/log4j/release-2.x#696 has test failures

2022-01-12 Thread Gary Gregory
Note that GitHub builds are green as are my local builds, both run twice as fast as this Jenkins instance which I guess must be oversubscribed, nothing bad about Jenkins itself of course. Gary On Wed, Jan 12, 2022, 13:03 Mr. Jenkins wrote: > *BUILD UNSTABLE* > Build URL > https://ci-builds.a

Re: [CI][UNSTABLE] Logging/log4j/release-2.x#696 has test failures

2022-01-12 Thread Ralph Goers
What is annoying is that the Jenkins builds seem to fail somewhere different every time and in places that I have never before seen fail. Ralph > On Jan 12, 2022, at 11:36 AM, Gary Gregory wrote: > > Note that GitHub builds are green as are my local builds, both run twice as > fast as this Je

Re: [logging-log4j2] 02/04: Our convention is to make test classes public.

2022-01-12 Thread Gary Gregory
I happy to stick with Junit 5 convertions once we drop Junit 4, which feels like a tediuous big job :-( Gafy On Wed, Jan 12, 2022, 13:33 Carter Kozak wrote: > +1 > > I prefer minimum visibility by default for the same reason I prefer to > make everything final by default: It gives us more freed

Re: [logging-log4j2] 02/04: Our convention is to make test classes public.

2022-01-12 Thread Matt Sicker
Most of the JUnit tests can be mechanically translated via IntelliJ (or other tools potentially). The tests that need manual migration are ones that use an expected exception in the @Test annotation and tests that use rules. I've already ported most of the JUnit 4 rules into equivalent JUnit 5 exte

Re: [logging-log4j2] 02/04: Our convention is to make test classes public.

2022-01-12 Thread Gary Gregory
Good to hear. I've stubbed my toes going from Junit rules 4 to Junit 5 extensions. I don't want to take the time to do that now though. Gary On Wed, Jan 12, 2022, 14:42 Matt Sicker wrote: > Most of the JUnit tests can be mechanically translated via IntelliJ > (or other tools potentially). The

Code Formatter?

2022-01-12 Thread Carter Kozak
Hi all, We've discussed the idea of using a code formatter before, I finally had a moment put up an example. Please take a look and provide feedback at your convenience :-) https://github.com/apache/logging-log4j2/pull/697 -ck