<headius>
enebo: ok, enxio and unixsocket added to the bundle...pulling the trigger
<headius>
it is done
<headius>
just waiting on the interwebs now
<camlow325>
Are there still plans to do a JRuby 1.7.27 release?
<headius>
camlow325: well, we're kinda dragging our heels on that but it could happen
<headius>
enebo: ?
<headius>
I think we have resigned ourselves to at least one more 1.7
<headius>
9.1.8 will be out this week and I guess we could look at 1.7.27 next
<baweaver>
Torvald's Law: No matter how often you warn of deprecation, at least one person will be irate with you after it happens.
<camlow325>
Cool, that would be great for us. We'd really like to move up to 9k soon. Have a few users really wanting 1.7.27 to drop so they can get the latest jruby-openssl bundled in it.
<baweaver>
Gates Corollary: ....and that one person will be an enterprise which will offer you large sums of money to make it work again.
<enebo>
yeah we always planned on one more point for 1.7 but have not gotten a lot of pressure for it
<GitHub153>
jruby/master a1c02e1 Thomas E. Enebo: Lengthen out for lack of determinism in this fixture. JRuby GC.start is no-op and we need to hope a real GC kicks in. 1000 seems to do it but we will make it much larger just in case
<enebo>
eregon: ^^^^^ This setup is non-deterministic but I helped improve our chances of it working
<enebo>
eregon: It should not affect MRI in that they obviously return in less than <100 calls
<GitHub189>
[jruby] headius pushed 3 new commits to master: https://git.io/vy3aO
<GitHub189>
jruby/master b31074d Charles Oliver Nutter: Fix global indy fail log to also mention uncacheable.
<GitHub189>
jruby/master a591b4a Charles Oliver Nutter: Add a blacklist of global globals that should not be cached....
<GitHub189>
jruby/master cf41d77 Charles Oliver Nutter: Add file and line and re-enable global indy fail log.
<GitHub199>
[jruby] headius closed issue #4508: $. never seems to change once JIT'd https://git.io/vyUph
<GitHub62>
[jruby] til opened issue #4515: Output of backticks call not returned but written to stdout instead https://git.io/vy3aH
<GitHub174>
jruby/master af0556b Thomas E. Enebo: new ruby/spec failure String#unpack(b) should be US-ASCII since it is literally only '1' or '0' chars
<nirvdrum>
headius: Thanks for the heads up.
<GitHub15>
[jruby] headius closed pull request #4511: Fix some issues with Fiddle (master...ds-fix-fiddle) https://git.io/vyku4
<GitHub68>
jruby/master a71de6b Daniel Smith: fix: add TYPE_PTRDIFF_T and friends
<GitHub68>
jruby/master 8f5b4c6 Daniel Smith: fix: incorrect invokation of __ffi_type__
<GitHub68>
jruby/master 6d129b7 Daniel Smith: fix: test fiddle helper not detecting platform properly
<jicksta>
what's the recommended way to instantiate a new ScriptingContainer / org.jruby.Ruby instance that is capable to loading a Rails 5 project into it? JRuby is embedded into a Scala project, so it doesn't have a gemset to use by default. I installed all the gems into `vendor/` with `bundle install --deployment` and `gem install -i bundler vendor/bundler`, but I keep getting stuck at requiring bundler. I assume once bundler is
<jicksta>
required properly, it'll read the .bundle/config properly and pull its gems from the `vendor/bundle` like normal with `--deployment`. Any tips for embedded JRuby + bundler + gems + Rails like this?
alex0ptr has quit [Remote host closed the connection]
joast has joined #jruby
alex0ptr has joined #jruby
alex0ptr has quit [Ping timeout: 240 seconds]
<lopex>
jicksta: if that helps, i've been a lot more successful using gem-maven-plugin
<lopex>
that not via container though
<lopex>
jicksta: though it could be usable from sbt via projectExtra xml or what it was called ?
<jicksta>
yeah was looking at the source to see if there were any tricks I could employ
<lopex>
did you try to minimize that in plain java an maven ?
<lopex>
via scripting container ?
<lopex>
*and
<jicksta>
what do you mean "try to minimize that"?
prasun has quit [Remote host closed the connection]
<jicksta>
oh, minimize tricks? I have tried a lot of stuff, setting the load paths explicitly, etc. i may need to setup more variables, like GEM_HOME et al
alex0ptr has joined #jruby
<GitHub101>
[jruby] ckpeter opened issue #4519: 64-bit immediate return to prompt on console interrupt before termination https://git.io/vysOi
alex0ptr has quit [Remote host closed the connection]
<lopex>
provide minimal java/maven/jruby scripting container issue like yours
alex0ptr has joined #jruby
<jicksta>
ah, yeah I have a sample project I could share
<lopex>
cool
<lopex>
since that would exclude ADFsbt issues
<lopex>
er
<lopex>
any
<lopex>
any sbt issues that is
<lopex>
jicksta: you can always plug pom parts in sbt config but I'm not sure what will they end up in maven phase cycle
<lopex>
classloader management and classpath management might be another issues
<lopex>
and ruby env vars is another topic :P
bbrowning is now known as bbrowning_away
<lopex>
er, LOAD_PATH
<lopex>
and in what classloader it will end up with
<lopex>
jicksta: would it be a problem to recreate that using maven ?
<lopex>
and call that scrupting container from plain java ?
<jicksta>
I'm much less familiar with Maven
<lopex>
*scripting
<lopex>
that no problem
<jicksta>
I can get it all the way to the point that it just can't find the bundler file to require...
<lopex>
sbt seems to have make-pom command
<jicksta>
yeah it does
<lopex>
can you try that ?
<jicksta>
I should add some setup instructions to the repo, but I was doing `bundle install --deployment` to install the gems to vendor/bundle, and `gem install -i bundler vendor/bundler` to install bundler specifically to vendor
<lopex>
my thinking is as follows
<jicksta>
so, installing JRuby is fine, I think the issue is more with the way that JRuby handles complex paths and such that a Rails project needs...
<lopex>
if we can exclude sbt/scala problems with minimal effort, then we could try maven in the same ways
<jicksta>
right now maven/sbt isn't touching the gem dependencies, but maybe it should through some sbt extension?
<jicksta>
when I instantiate the org.jruby.Ruby instance, it can execute files and such. just doesn't have a gemset or any gems installed
<lopex>
that's what we need to figure, but using maven would make things easier wrt minimizing the case
<jicksta>
what's your theory about maven vs sbt causing an issue? that it isn't installing an setting up jruby and the classpath for the scala process properly?
<lopex>
absolutely no idea
<lopex>
but sbt apart from using ivy is a totally separate tool
<jicksta>
there's apparently some setup that I need to deal with to configure the JRuby environment to boot up a Rails app. I didn't want to use warbler because I want to invoke Ruby methods from Scala, passing in Scala case classes
<lopex>
from now on, it's the easiest step to get rid of anything that can make a problem
<jicksta>
i.e. don't want a server. just the files required in production mode
<lopex>
sure
<lopex>
jruby/rails is a complex beast so you want to isolate that as much as possible
<lopex>
and you have minimal sbt
<lopex>
if build conf wuld be complex that wuld be a different story
<lopex>
of course I can be mistaken anyways, it's just what my experience dictates me
<jicksta>
I think I need to understand what warbler is doing more... it looks like it can generate a gemjar. don't know how to add that to the classpath though
<lopex>
also, I'd try looking up some sbt classpath management docs
<jicksta>
lopex: do you know anything about jruby's use of .jar files in its LOAD_PATH / Classpath ?
<lopex>
jicksta: only that jruby uses it;s own classloader to load anything
<lopex>
and it can cause problems
<jicksta>
yeah, i've seen the JRubyClassLoader IIRC
<lopex>
since system classloader wont see thoses clases
<lopex>
it's a problem with jdbc drivers for example
<lopex>
unless it's been solved nowadays
<lopex>
that I dont know
<lopex>
yeah, dragons everywhere
<lopex>
actually, I havent tried child first under jruby
<lopex>
jicksta: it;s really simple there;s classloader tree, so every class loaded by given classloader wont be visible by child classloader loaded classes
<lopex>
given parent-first
<lopex>
but you can reverse that
<jicksta>
ok. trying to get the gemjar generated with warbler... not seeing any outputted file tho. trying to add it manually to the jruby class path
<lopex>
jicksta: if any of all of that fails, you can always create jruby runtime directly
<lopex>
but if you can provide minimal maven case that would be great
<lopex>
jicksta: and by "directly" I mean you can force any jruby runtime propes like the LOAD_PATH