Displaying #maven-dev/2017-02-05.log:

Sun Feb 5 05:43:58 2017  olamy:Joined the channel
Sun Feb 5 13:53:40 2017  tibor_:Joined the channel
Sun Feb 5 15:45:13 2017  britter:Joined the channel
Sun Feb 5 15:45:36 2017  britter:hello @tibor_
Sun Feb 5 15:48:28 2017  tibor_:britter: Hi Benedikt
Sun Feb 5 15:48:35 2017  tibor_:What's up
Sun Feb 5 15:48:35 2017  britter:How are you?
Sun Feb 5 15:48:40 2017  tibor_::)
Sun Feb 5 15:48:44 2017  tibor_:fine now
Sun Feb 5 15:48:53 2017  britter:yeah, worked on some ASF stuff for the first time in ages today
Sun Feb 5 15:48:56 2017  tibor_:so what can we do?
Sun Feb 5 15:49:16 2017  britter:I'm not sure
Sun Feb 5 15:49:28 2017  tibor_:a talk with Marc as well?
Sun Feb 5 15:49:29 2017  britter:I'm pretty much stuck currently.
Sun Feb 5 15:49:52 2017  britter:I can ask whether he has some time. But maybe it's better to do a video call then.
Sun Feb 5 15:50:08 2017  britter:three people typing = chaos :)
Sun Feb 5 15:50:24 2017  tibor_:not bad idea
Sun Feb 5 15:51:09 2017  tibor_:would you talk with him and manage it?
Sun Feb 5 15:51:37 2017  britter:I thought it would be a good think if the three of us met in a video call.
Sun Feb 5 15:51:51 2017  tibor_:definitely
Sun Feb 5 15:52:00 2017  britter:but maybe the two of us discuss the current state first and then we get back to Marc
Sun Feb 5 15:52:16 2017  tibor_:makes sense
Sun Feb 5 15:52:36 2017  britter:So there are basically two things blocking me currently (aside from having no time ;-)
Sun Feb 5 15:53:07 2017  tibor_:what kind of?
Sun Feb 5 15:53:10 2017  britter:1. Setting up a integration test environement which runs the same tests against junit 4 and 5.
Sun Feb 5 15:53:46 2017  britter:2. Not having the code of the junit 5 surefire adapter in the surefire repo. There is no way for me to fix things on the go, since we depend on their artifact
Sun Feb 5 15:54:30 2017  tibor_:it's obious to me that my work is too slow to help you
Sun Feb 5 15:54:38 2017  britter:2. is pretty easy to solve. They have relicensed their adapter under terms of ALv2 and they would like to donate it to maven-surefire. This is something I could manage with Marc.
Sun Feb 5 15:54:53 2017  tibor_:my propose was to include junit5 in 3.0 RC1
Sun Feb 5 15:55:06 2017  britter:yes. But we need the code to make that happen.
Sun Feb 5 15:55:58 2017  tibor_:For your information what happened in ASF Mavenn
Sun Feb 5 15:56:02 2017  britter:from your PoV would it be an option to accept a code donation of the adapter code?
Sun Feb 5 15:56:22 2017  britter:I'm following the ML so I kind of know what happend. nasty stuff going on in maven land :-)
Sun Feb 5 15:56:38 2017  britter:but go ahead. It's interesting to hear your PoV.
Sun Feb 5 15:56:38 2017  tibor_:We reverted 11 commits in surefire, and today I brought few of them to branch SUREFIRE-1322
Sun Feb 5 15:57:30 2017  tibor_::)
Sun Feb 5 15:58:03 2017  tibor_:the next would be two branches and one more pending on my disk - finished
Sun Feb 5 15:58:30 2017  tibor_:I need Herve to accept the branches and then we can go on with release
Sun Feb 5 15:58:49 2017  tibor_:Regarding 3.0, the branch already exists
Sun Feb 5 15:59:08 2017  tibor_:it is almost done but one bug is there and must be fixed
Sun Feb 5 15:59:11 2017  britter:Why is Herve involved in surefire?!
Sun Feb 5 15:59:55 2017  tibor_:Herve is here long time and he told me that he was open to participate in the review.
Sun Feb 5 16:00:16 2017  britter:okay
Sun Feb 5 16:00:37 2017  britter:i understand
Sun Feb 5 16:00:40 2017  tibor_:It's good because we can prevent from chaos in master and prevent from next reverts.
Sun Feb 5 16:01:11 2017  britter:GitHub has this nice feature we're you can protect master branch vom receiving direct commits.
Sun Feb 5 16:01:19 2017  britter:would be good to have this in surefire.
Sun Feb 5 16:02:33 2017  tibor_:This slows down but the latency cause from miserable feeling when you give it up is higher than without using branches.
Sun Feb 5 16:03:43 2017  tibor_:what is that feature? I do not know it.
Sun Feb 5 16:04:45 2017  tibor_:Herve told me that he is not using GitHub nothing but Git in IDE. I guess the rest of the team has similar experiences.
Sun Feb 5 16:07:54 2017  britter:tibor_ my girl friend just came home. I'll be back in like 15 minutes.
Sun Feb 5 16:08:38 2017  tibor_::) enjoy it. I would go for longer talk than 15 min.
Sun Feb 5 16:08:48 2017  tibor_:sorry walk
Sun Feb 5 16:08:53 2017  tibor_:not talk :)
Sun Feb 5 16:24:12 2017  britter:okay, here I am again.
Sun Feb 5 16:24:17 2017  britter:let's decide on the next steps.
Sun Feb 5 16:26:39 2017  britter:tibor_ how about we push the import of the surefire-provider code from the JUnit project next?
Sun Feb 5 16:27:02 2017  britter:once we have that, we can take some more time to think about how we can write tests for that.
Sun Feb 5 16:45:26 2017  tibor_:ok, I am back.
Sun Feb 5 16:46:29 2017  tibor_:We should make video call to talk about tests. This is too slow chat, and talking is so natural for us.
Sun Feb 5 16:46:40 2017  tibor_:Maybe one remark.
Sun Feb 5 16:47:17 2017  tibor_:After release of 2.189.2, we can directly push 3.0 RC1 to master and develop 2.19.3 which suppoorts JDK 9 with only one issue.
Sun Feb 5 16:47:39 2017  tibor_:And this way everything is ready for then.
Sun Feb 5 16:48:01 2017  tibor_:I can help writing tests as well in 3.0
Sun Feb 5 17:02:14 2017  britter:tibor_ pretty async conversation :-)
Sun Feb 5 17:03:05 2017  tibor_:ok, I am here
Sun Feb 5 17:03:13 2017  tibor_:let's continue
Sun Feb 5 17:03:14 2017  britter:okay, do you want to have the surefire-provider code for 3.0 or for 2.19.3 ?
Sun Feb 5 17:04:07 2017  tibor_:3.0 been in master, and 2.19.3 in branch in parallel because this would have only jdk9 support which is one issue
Sun Feb 5 17:04:28 2017  britter:okay so the provider code should go to master, right?
Sun Feb 5 17:04:30 2017  tibor_:I don't want you to wait too long
Sun Feb 5 17:05:20 2017  tibor_:yes, since RC versions are good thing to let the clients to try out
Sun Feb 5 17:05:55 2017  tibor_:WDYT?
Sun Feb 5 17:06:47 2017  britter:yes
Sun Feb 5 17:06:51 2017  britter:excellent idea
Sun Feb 5 17:07:28 2017  britter:okay, so I talk to Marc then, that he should prepare a PR against master containing the provider code. Would that work for you?
Sun Feb 5 17:08:00 2017  britter:Once we have the PR in place, we can discuss in a video call how we want to go on with the tests.
Sun Feb 5 17:09:21 2017  tibor_:britter: you know, 3.0 changes everyting like removing some config params and introducing new once, introducing extensions, etc. The people may shout if I just apply another system property for skipTests.
Sun Feb 5 17:10:11 2017  britter:I don't think I understand what you're implying here :-)
Sun Feb 5 17:10:20 2017  britter:So you don't want the PR against master?
Sun Feb 5 17:10:29 2017  tibor_:Therefore it is better to prepare the users and their systems to adapt for breaking the backwards compatibility. JUnit5 is fitting here because the changes take some time same as JUnit5.
Sun Feb 5 17:12:11 2017  tibor_:PR for master is better than for branch. But PR is the only way for you since you are not Maven committer.
Sun Feb 5 17:13:02 2017  tibor_:Look we always found consensus, we both. There is no reason for opposite.
Sun Feb 5 17:14:04 2017  britter:okay tibor_! I'm talking to Marc to prepare the PR.
Sun Feb 5 17:14:32 2017  britter:After that, we get together in the video call and discuss how I can help prepare more tests for the provider code.
Sun Feb 5 17:17:01 2017  tibor_:I see power in you and without you the JUnit5 would not be in Surefire. I only want to find the way to have it soon in release and combine the original plans of 3.0 with JUnit5 and anf align the time for both and this seems to be doable in 3.0 RC1.
Sun Feb 5 17:17:35 2017  britter:Yes, we can make that together.
Sun Feb 5 17:18:30 2017  tibor_:finilizing junit 5 and surefire extensions may take approximately the same time and in both cases the users may react and prepare as well.
Sun Feb 5 17:22:50 2017  tibor_:From this, since developing two big features, one feature should not wait for the other feature. This of course requires fair collaboration, because without it one is blocked, but opposite is valid so that regular work on both means that we fairly end up at the same time and we will be happy.
Sun Feb 5 17:26:19 2017  britter:tibor_ if we can get the junit 5 provider code in a shape we're we want to give it to users, we can use feature toggles to deactivate it, so that we have more time to get it ready
Sun Feb 5 17:27:30 2017  tibor_:Yes, exactly.
Sun Feb 5 17:29:50 2017  tibor_:On the other hand, if we stay in branch not doing a release with JUnit5 preliminary solution, the users would not know and would not react. If we added to RC1 then we would have a chance to get one more developer from users.
Sun Feb 5 17:31:26 2017  britter:yes.
Sun Feb 5 17:32:03 2017  britter:I recommend working with the junit 5 code on master. Doing it on a separate branch leads to late integration. It will cause problems and pain during development.
Sun Feb 5 17:32:19 2017  tibor_:sure
Sun Feb 5 17:33:33 2017  britter:okay, I'll talk to Marc and get back to you soon.
Sun Feb 5 17:33:42 2017  tibor_:feel free
Sun Feb 5 17:36:38 2017  britter:Joined the channel
Sun Feb 5 17:41:42 2017  britter:Joined the channel
Sun Feb 5 17:42:05 2017  britter:Joined the channel
Sun Feb 5 18:27:21 2017  britter:Joined the channel
Sun Feb 5 18:50:11 2017  britter:Joined the channel
Sun Feb 5 19:03:04 2017  britter:tibor_ here is the code donation: https://issues.apache.org/jira/browse/SUREFIRE-1330 I will start work on that shortly. I looked at master branch but is has version 2.19.2-SNAPSHOT. When will it be bumped to 3.0?
Sun Feb 5 19:03:40 2017  tibor_:3.0 or 3.0 RC1?
Sun Feb 5 19:04:09 2017  tibor_:3.0 RC1 much sooner I would say
Sun Feb 5 19:04:21 2017  britter:do you put RC1 into <version> tag?
Sun Feb 5 19:04:43 2017  tibor_:since a lot of migration from Maven 2 to Maven 3 has been done one year agon in branch 3.0-rc1
Sun Feb 5 19:04:57 2017  tibor_:I think so
Sun Feb 5 19:05:15 2017  tibor_:you change change the version in Jira ticket
Sun Feb 5 19:05:29 2017  britter:to 3.0-RC1?
Sun Feb 5 19:05:33 2017  tibor_:yes
Sun Feb 5 19:06:00 2017  britter:like this?
Sun Feb 5 19:06:26 2017  britter:I think it should be fixed for 3.0-RC1 and 3.0 so people looking for 3.0 see that it is included in 3.0
Sun Feb 5 19:07:17 2017  tibor_:I use to have only one version. What impact two version would have in Jira? Reopening the same ticket?
Sun Feb 5 19:07:45 2017  britter:okay, setting to 3.0-RC1 only
Sun Feb 5 19:08:23 2017  britter:so when I work on that, I base my work on the 3.0-rc1 branch until you merge it to master.
Sun Feb 5 19:12:38 2017  tibor_:We can put commits to 3.0-rc1 branch but the problem with the branch is that it contains a lot of mess. I will cherry pick those changes in master after we released 2.19.2.
Sun Feb 5 19:13:11 2017  tibor_:So you can commit there but please commit one bit change because this simplifies my work then.
Sun Feb 5 19:14:18 2017  tibor_:Especially if you want to continue in AbstractSurefireMojo class, it is good to commit in 3.0-rc1 and because this class changed a lot after migrated from Maven 2 to 3.0.
Sun Feb 5 19:14:31 2017  tibor_:The providers API did not change.
Sun Feb 5 19:15:51 2017  tibor_:Remember that Surefire141*IT fails because transitive dependencies from junit runner was supressed as a consequence of migrating to Maven 3.0. It is a bug.
Sun Feb 5 19:16:11 2017  tibor_:This needs to be fixed and the build will be okay.
Sun Feb 5 19:17:51 2017  tibor_:Good news. Herbe replied on my email Although he is skiing this week, he will take a look in my branch. Meanwhile I will continue with other branch and we will finish 2.19.2 soon.
Sun Feb 5 19:18:05 2017  tibor_:sorry Herve, my typo.
Sun Feb 5 19:20:47 2017  tibor_:britter: do you think we will be able to merge the branch junit5 back to master after 2.19.2 release?
Sun Feb 5 19:21:40 2017  tibor_:Maybe we should rebase the branch on the top of master's head.