lucf117 has quit [Remote host closed the connection]
AdamGordonBell[m has quit [Quit: node-irc says goodbye]
<headius> good morning!
<basshelal[m]> good morning
<headius> taking a moment to check our pass rates on JDK16 since that release made illegal access "deny" by default
<headius> so far looks fine
<basshelal[m]> I tried to move JFFI to JUnit5 but that was embarrassing, might be hard for me but might not even be as useful since it's more C than Java anyway
<basshelal[m]> JNR-POSIX is tricky too I think but I'm a little exhausted for it
<headius> yeah the jffi stuff is a little more involved
<basshelal[m]> I posted a new issue on JNR-FFI regarding a benchmarking suite, let me know what you think
<headius> ok
<headius> enebo: are you able to run our rake tasks on master?
<headius> I am getting a weird error from the ant stuff, ant.path does not exist so the build doesn't load
<enebo[m]> Did anything change since yesterday?
<headius> I don't think so but I had not run anything since back from vacay
<headius> it appears to be green on travis so I assume it is just my env
<enebo[m]> I had run spec:ruby:fast yesterday afternoon and it was a branch I just made
<enebo[m]> there were some commits during the day yesterday so I am rebuilding
<enebo[m]> I am using Java 8 too
<headius> hmm ok
<headius> NoMethodError: undefined method `path' for nil:NilClass
<headius> Did you mean? mkpath
<headius> /Users/headius/projects/jruby/rakelib/commands.rake:10:in `initialize_paths'
<enebo[m]> any particular one you want?
<enebo[m]> any at all
<headius> NoMethodError: undefined method `path' for nil:NilClass
<headius> Did you mean? mkpath
<headius> /Users/headius/projects/jruby/rakelib/commands.rake:10:in `initialize_paths'
<headius> any at all really but I was running spec"ruby"fast
<headius> ::
<enebo[m]> yeah. running now
<headius> ok something is screwed up here then
<enebo[m]> I thought someone said something about later Javas and ant not working but this would not appear to be that
<headius> I think I found it
<headius> I had a rogue ant.jar in lib/ probably from my experiments trying to fully modularize
<headius> so it was loading that instead of the ext
<headius> might need to scrub this env soon
<headius> I wanted to push a PR that enables some suites on JDK16 since that one denies all illegal access by default
<headius> good to know before release that things pass the same on 16 as they do currently on 14
<headius> I pushed 14 to 16 move for travis which should be basically a no-op, but I will also add JDK16 runs to GHA
<enebo[m]> cool
<headius> there were a couple other tweaks for that MacOS no-JAVA_HOME thing last night but it should be fixed everywhere now
<headius> pushed a new launcher with the change
<headius> windows could use a new exe but it did not touch any windows code paths
<headius> hmmm the GHA builds will be a problem because the Ruby plugins still run with an old version of JRuby that does not behave well
<headius> seems to be failing booting up some of the dist testing stuff
<headius> I think I will punt on the GHA builds until 9.3 is out, so we can move the plugins to that release
<headius> not sure how else to make it work other than running maven with a bunch of module flags that might taint results
<headius> enebo: we should start talking about what remaining items to push to 9.3.1 so we can get this thing 9.3 out by end of July
<headius> like getting down to just critical items: autoload regression, memory/stack problems in JI
<headius> or rather perhaps we should be talking about minimum remaining work to get 9.3 out
<headius> we know it will require many update releases once people actually start testing it anywa
<headius> enebo: I told that DirkMahler to do the JarCache improvements on 9.2 since that will be the one asciidoctor and other key embedders use for quite a while still
<headius> in service of that I backported a commit from last year that fixed the weak/soft cache issue: https://github.com/jruby/jruby/pull/6734
architecture[m] has joined #jruby
<enebo[m]> headius: yeah looking through the last 54 issues most of them seem puntable to me. I will fix the case/when logic. There is some coverageish stuff we are really lacking on but that is partially a problem since 2.5
<headius> yeah punt away
<headius> 9.3.1 or 9.4 for stretch goals
<headius> coverage would be a good 9.3.1 job
<headius> or 9.3.x
<headius> we have only gotten a few reports about it so I am not sure how heavily it is used
<enebo[m]> You opened a lot of it would be nice bugs like https://github.com/jruby/jruby/issues/5862
<enebo[m]> You may want to take a pass over these and mark milestones on this class of issues that you opened
<headius> ok
<headius> pretty much all of my would be nice will be 9.4
<headius> enebo: I did a first pass for 9.4 moves, down to 43 open
<headius> we need to review PRs also... there are some nice to haves and a bunch of stale red PRs
<headius> you have one for nanosecond time for example that is green
<headius> enebo: I will punt all coverage related stuff to 9.3.1
<headius> that requires work from both your end and mine so it clearly won't happen for .0 but we might be able to get it working before 9.4
<enebo[m]> sure I think 9.4 is fine for it since I barely even remember what it is and I have had doubts about it ever since :)
<headius> mostly things like branch and code block coverage etc
<headius> just finer-grained tracking, not sure what is different in CRuby to support that tracking
<enebo[m]> those for sure I don't think are needed we can put in notes we plan on working on them later
<enebo[m]> release notes on what is not there
<enebo[m]> When I started looking at those two issues I could see we may need more metadata about code than what we are capturing to know it is in a method or a block or whatnot
<enebo[m]> branching I did not even think about
<headius> down to 33 open I want to review in detail
<headius> 14 PRs
<headius> enebo: I'm moving on to triaging the remaining 9.3 items to see what is still an issue and what can be punted and what needs to be finished
<enebo[m]> ok
<headius> kares: I have noticed after some partial builds locally, like just doing 'mvn', lib/ruby/stdlib/rubygems.rb gets wiped out
<headius> not sure why that is or if it is just my env but could your logic to install rubygems at build time cause this
<headius> ?
<headius> a clean rebuild fixes it but not sure why this file disappears