Displaying #maven-dev/2017-01-03.log:

Tue Jan 3 09:48:24 2017  tibor_:Joined the channel
Tue Jan 3 11:08:12 2017  rfscholte:Joined the channel
Tue Jan 3 14:24:23 2017  rfscholte:stephenc: ping
Tue Jan 3 14:24:33 2017  stephenc:yep
Tue Jan 3 14:24:47 2017  rfscholte:which issues are missing JIRAs?
Tue Jan 3 14:25:16 2017  rfscholte:the commits you are mentioning
Tue Jan 3 14:25:20 2017  stephenc:MNG-xxxx WONTFIX Add a ProjectArtifactsCache similar to
Tue Jan 3 14:25:20 2017  stephenc: PluginArtifactsCache
Tue Jan 3 14:25:20 2017  stephenc:MNG-xxxx WONTFIX added core extensions documentation
Tue Jan 3 14:25:20 2017  stephenc:MNG-xxxx WONTFIX added slf4j-api and jansi dependencies
Tue Jan 3 14:25:20 2017  stephenc:MNG-xxxx WONTFIX Added some docs in CLIReporting Utils
Tue Jan 3 14:25:22 2017  stephenc:MNG-xxxx WONTFIX Avoid implicit conversions of int to long
Tue Jan 3 14:25:25 2017  stephenc:MNG-xxxx WONTFIX Cleaned up code using the ComparableVersion
Tue Jan 3 14:25:27 2017  stephenc: class prevents warnings in IDE about raw
Tue Jan 3 14:25:30 2017  stephenc: types.
Tue Jan 3 14:25:32 2017  stephenc:MNG-xxxx WONTFIX cleanup previous installation before installing
Tue Jan 3 14:25:35 2017  stephenc: the new one
Tue Jan 3 14:25:37 2017  stephenc:MNG-xxxx WONTFIX Code polishing o Replaced
Tue Jan 3 14:25:40 2017  stephenc: Collections.unmodifiableList() with
Tue Jan 3 14:25:42 2017  stephenc: google ImmutableSet to keep the original Set
Tue Jan 3 14:25:45 2017  stephenc: type.
Tue Jan 3 14:25:47 2017  stephenc:MNG-xxxx WONTFIX Code polishing o Replaced static initializer
Tue Jan 3 14:25:50 2017  stephenc: block wi
Tue Jan 3 14:25:52 2017  stephenc:etc
Tue Jan 3 14:25:55 2017  stephenc:there were a pile of commits that do not have an associated JIRA issue... unsure how many need one
Tue Jan 3 14:26:03 2017  stephenc:unsure how many we want to keep
Tue Jan 3 14:26:18 2017  stephenc:but we need an assessment / decision on all of them
Tue Jan 3 14:26:55 2017  stephenc:some of those are clearly the "using master as a development branch" style, e.g. the 3rd one I pasted is MNG-3507
Tue Jan 3 14:27:32 2017  rfscholte:do you expect my opinion on all of these issues? also why i didn't selected other issues from the available list?
Tue Jan 3 14:27:54 2017  stephenc:I expect *somebody* to assist in assessing those
Tue Jan 3 14:28:03 2017  stephenc:I do not expect any specific person
Tue Jan 3 14:28:04 2017  rfscholte::) no problem
Tue Jan 3 14:28:53 2017  stephenc:When we get to the point of actually having the vote passed and the revert in place, then we can see what hasn't been considered worthy of a JIRA
Tue Jan 3 14:29:05 2017  stephenc:and what has been lumped in with existing JIRAs
Tue Jan 3 14:29:22 2017  stephenc:and if nobody else has helped out, well as release manager I will have to just make a call
Tue Jan 3 14:29:40 2017  stephenc:(by default call being "it's out")
Tue Jan 3 14:30:11 2017  stephenc:@rfscholte you see me making a call is easy
Tue Jan 3 14:30:21 2017  stephenc:but it may not be best for the code ;-)
Tue Jan 3 14:30:41 2017  stephenc:OTOH I will be doing a 3.5.1 six weeks later, so it shouldn't be a major harm!
Tue Jan 3 14:31:29 2017  rfscholte:I'm not the git-guru yet, especially in these special situations...
Tue Jan 3 14:31:52 2017  stephenc:no worries. I can help!
Tue Jan 3 14:32:19 2017  stephenc:I don't call myself a git guru, but I am self confident
Tue Jan 3 14:32:25 2017  stephenc:;-)
Tue Jan 3 14:33:26 2017  rfscholte:great :)
Tue Jan 3 14:36:18 2017  stephenc:well I'm at "has successfully had a feature request added to Git" level of usage
Tue Jan 3 15:21:07 2017  stephenc:@aheritier @rfscholte w00t https://builds.apache.org/view/Maven/job/maven-jenkinsfile/job/master/57/console
Tue Jan 3 15:21:34 2017  stephenc:now have integration tests running on nix+j7, nix+j8, win+j7 and win+j8 in parallel all as part of the same build job
Tue Jan 3 15:22:03 2017  stephenc:any using multibranch, so if you create a new branch you automatically get a build job created for you
Tue Jan 3 15:22:31 2017  stephenc:(still have to sort out replicating this for PRs, but now that master has a working Jenkinsfile we can use trusted revision for that)
Tue Jan 3 15:23:09 2017  stephenc:and then have to sort out the final bit - which is getting infra to have builds.apache.org use latest plugins so we get immediate response to git commits
Tue Jan 3 15:38:35 2017  rfscholte:nice!
Tue Jan 3 15:39:20 2017  stephenc:dear god the windows tests are an order of magnitude slower
Tue Jan 3 15:40:16 2017  rfscholte:"you need to use 'install' with the integration tests not 'verify'" this is a bad sign, would be nice if that could be fixed in the future
Tue Jan 3 15:40:40 2017  stephenc:@rfscholte, yep... using mrm would fix it
Tue Jan 3 15:40:51 2017  stephenc:as mrm can resolve from the reactor
Tue Jan 3 15:41:35 2017  rfscholte:I don't think many know of the power you get with mrm
Tue Jan 3 15:41:36 2017  stephenc:but for now I just want to get the tests running easily for everyone without needing to run around managing jobs
Tue Jan 3 15:41:52 2017  stephenc:+1
Tue Jan 3 15:42:03 2017  stephenc:it was one of my better inventions ;-)
Tue Jan 3 15:42:25 2017  rfscholte:I totally agree :D
Tue Jan 3 15:46:30 2017  jasonvanzyl:Joined the channel
Tue Jan 3 15:47:09 2017  stephenc:@jasonvanzyl so I want to act as release manager for 3.5.0... can you share any magic that you do for a core release?
Tue Jan 3 15:47:39 2017  jasonvanzyl:sure, let me try and find all my stuff. amounts to a couple tools
Tue Jan 3 15:47:47 2017  jasonvanzyl:generally that i just run from the IDE
Tue Jan 3 15:47:58 2017  jasonvanzyl:it's been a while so i need to refresh my memory
Tue Jan 3 15:48:21 2017  stephenc:no rush... I still have to get a big reset of master done and then build back all the stuff we want in
Tue Jan 3 15:48:27 2017  jasonvanzyl:https://github.com/jvanzyl/source-release-validator
Tue Jan 3 15:48:55 2017  jasonvanzyl:i think for us if you just do a hard reset and force push we can all clone again
Tue Jan 3 15:48:59 2017  stephenc:@jasonvanzyl: can you put the braindump in a mail to me when you have a chance
Tue Jan 3 15:49:00 2017  jasonvanzyl:i don't think that's a huge deal
Tue Jan 3 15:49:33 2017  jasonvanzyl:i am somewhat empathetic with christian as i have done what he's done in the past when i was his age, but we didn't have nearly so many users
Tue Jan 3 15:49:33 2017  stephenc:jasonvanzyl: yes but infra has a lock on master to prevent force pushes, so I need to raise an infra ticket
Tue Jan 3 15:50:12 2017  jasonvanzyl:as far as i can tell most things work, igor and i both tried some things. what is disconcerting is mangling of master and his decisions on deciding what's "right"
Tue Jan 3 15:50:18 2017  stephenc:jasonvanzyl: I am somewhat of the same opinion. He has great evergy and his passion is coming from the right place
Tue Jan 3 15:50:22 2017  jasonvanzyl:i'm sure he means well
Tue Jan 3 15:50:42 2017  jasonvanzyl:right, and i've done the same things in a stupid manner so i get it
Tue Jan 3 15:50:48 2017  stephenc:... but... we have let him down by not having the right processes in place
Tue Jan 3 15:51:24 2017  jasonvanzyl:maybe, i don't want to be over burdened by process and i don't know how you convey to someone younger that burning things to the ground isn't productive
Tue Jan 3 15:51:44 2017  jasonvanzyl:but how do you teach that, we all think we can do anything until we hit 35
Tue Jan 3 15:51:57 2017  stephenc:I'm just trying to be magnanimous in throwing blame around
Tue Jan 3 15:52:08 2017  jasonvanzyl:i'd prefer he just be reasonable and deciding arbitrarily that something works this way and not that is not reasonable given the users we have
Tue Jan 3 15:52:25 2017  stephenc:Jason... I am going to have to screenshot that and post it on my wall
Tue Jan 3 15:52:40 2017  jasonvanzyl:he's well intentioned, we probably had some lack of guidance and he doesn't seem inclined to listen. shit happens.
Tue Jan 3 15:52:57 2017  stephenc:@ASFBot: pointer
Tue Jan 3 15:53:26 2017  jasonvanzyl:but ultimately if it starts breaking 10% of our users that's not good and changing tests, ITs and code at the same time is not good
Tue Jan 3 15:54:01 2017  jasonvanzyl:Igor and I certainly took some liberties but we never changed the ITs, made sure they worked, wrote interfaces for for new implementations where the default provided the existing behavior and you opt into the new behavior
Tue Jan 3 15:54:22 2017  jasonvanzyl:which is a lot of work, but that's how we started doing it because we needed something different but are cognizant of not breaking people
Tue Jan 3 15:54:27 2017  jasonvanzyl:he didn't seem to want to listent
Tue Jan 3 15:54:29 2017  stephenc:yep: I know
Tue Jan 3 15:55:03 2017  stephenc:look, releasing software that gets used by many thousands of developers is a different skillset from other development
Tue Jan 3 15:55:27 2017  stephenc:and retaining backwards compatibility while doing all that is another skill set again
Tue Jan 3 15:55:36 2017  jasonvanzyl:so one simple rule can be ITs are immutable
Tue Jan 3 15:55:46 2017  stephenc:I think existing ITs are RTC
Tue Jan 3 15:55:50 2017  jasonvanzyl:if we made a mistake, at worst you change the range it runs in and write another IT
Tue Jan 3 15:56:00 2017  rfscholte:+1
Tue Jan 3 15:56:05 2017  jasonvanzyl:if it's purely additive it's fine
Tue Jan 3 15:56:09 2017  jasonvanzyl:he started changing ITs
Tue Jan 3 15:56:15 2017  jasonvanzyl:which is just not wise
Tue Jan 3 15:56:19 2017  stephenc:new ITs is CTR IMHO
Tue Jan 3 15:56:29 2017  jasonvanzyl:sure, that's reasonable
Tue Jan 3 15:56:34 2017  jasonvanzyl:https://github.com/jvanzyl/source-release-validator/blob/master/src/main/java/io/tesla/sra/SourceReleaseAnalyzer.java#L288
Tue Jan 3 15:56:50 2017  jasonvanzyl:i'll write some notes but basically i do the release to nexus and update the values in the main there
Tue Jan 3 15:56:51 2017  stephenc:I think saying existing ITs are immutable is a bridge too far, there may be legitimate cases where we identify that an integration test was incorrectly written
Tue Jan 3 15:57:18 2017  jasonvanzyl:sure, but then mark it only to run in the version it failed in and copy it and fix it
Tue Jan 3 15:57:30 2017  jasonvanzyl:to provide a more clear history of the mistake
Tue Jan 3 15:57:41 2017  stephenc:that's for one class of mistake
Tue Jan 3 15:58:02 2017  jasonvanzyl:basically append only
Tue Jan 3 15:58:05 2017  stephenc:I'm saying say we have an integration test that we think is testing something, but then we find a bug in the integration test
Tue Jan 3 15:58:06 2017  jasonvanzyl:like a journal
Tue Jan 3 15:58:26 2017  stephenc:we shouldn't be keeping the broken test that give a false positive
Tue Jan 3 15:58:51 2017  stephenc:the test that gives a false negative also should be fixed
Tue Jan 3 15:59:03 2017  stephenc:but it needs to be a problem with the test
Tue Jan 3 15:59:15 2017  jasonvanzyl:i would and either just disable it or limit it to a range as if something was released which that IT allowed a certain behavior it's in the release
Tue Jan 3 15:59:20 2017  stephenc:if it is a case that the test is no longer valid, then that is a case where we adjust the range
Tue Jan 3 15:59:30 2017  jasonvanzyl:if the IT is actually just broken i still think it should be capture, maybe git is enough
Tue Jan 3 15:59:37 2017  jasonvanzyl:but i believe this is extremely rare
Tue Jan 3 15:59:44 2017  jasonvanzyl:that the IT is itself broken
Tue Jan 3 16:00:07 2017  stephenc:I am saying if the test never actually worked correctly, we should fix it. There was exactlt one case that I am aware of for an IT where the IT was actually broken and giving false results
Tue Jan 3 16:00:19 2017  jasonvanzyl:we should definitely fix it
Tue Jan 3 16:00:24 2017  stephenc:I did fix that one
Tue Jan 3 16:00:45 2017  stephenc:my point is that there are valid cases where an IT may need to be changed... for those cases it should be RTC
Tue Jan 3 16:00:59 2017  stephenc:new ITs, CTR... range changes, RTC
Tue Jan 3 16:01:27 2017  stephenc:changes to the spec of maven should be covered by range changes
Tue Jan 3 16:25:50 2017  aheritier:nice work stephenc with the pipeline job
Tue Jan 3 16:26:16 2017  stephenc:aheritier... not quite there yet... but getting close
Tue Jan 3 16:26:26 2017  aheritier:yes
Tue Jan 3 18:04:25 2017  olamy:Joined the channel
Tue Jan 3 18:27:42 2017  hboutemy:Joined the channel
Tue Jan 3 20:02:40 2017  olamy:Joined the channel
Tue Jan 3 20:02:54 2017  stephenc:@olamy are you ok with the big reset?
Tue Jan 3 20:14:06 2017  stephenc_:Joined the channel
Tue Jan 3 20:39:39 2017  jasonvanzyl_:Joined the channel
Tue Jan 3 22:48:13 2017  jasonvanzyl:Joined the channel

Comments