<donv[m]>
headius: I just started using Java 11 from AdoptJDK and it works well, but I am getting the warning about SecurityHelper. Should that have been fixed in 9.2.11.1 ?
<headius[m]>
Do you have JAVA_HOME set?
<donv[m]>
Maybe not. Do I need to?
<donv[m]>
headius:
<donv[m]>
headius:
<donv[m]>
I tried running with JAVA_HOME set, and the warning does not appear. Why is that?
<headius[m]>
The launchers will not add module flags unless they can determine it will run on a Java module enabled JVM, which it does by looking for jmods dir
<headius[m]>
I would like a better way but launching the jvm even to check version takes over a tenth of a second
<donv[m]>
headius: Got an issue introduced by RuboCop: `before_validation{|e| e.my_method}` is converted to `before_validation(&:my_method)` and fails with "ArgumentError no receiver given"
<donv[m]>
Checking with MRI now.
<donv[m]>
Same with MRI, so not a JRuby issue. Do you understand how the two syntaxes are different?
lucasb has joined #jruby
subbu|away is now known as subbu
<rdubya[m]>
does anybody know why the alpine builds of the docker image don't get pushed to docker hub? It appears the dockerfiles get updated and they build fine for me locally, but the tags aren't on docker hub
xardion has quit [Remote host closed the connection]
xardion has joined #jruby
shellac has quit [Ping timeout: 240 seconds]
sagax has quit [Remote host closed the connection]
sagax has joined #jruby
<rdubya[m]>
headius: enebo Do either of you know of any issues using Java 13 with JRuby 9.2? We're looking to update from Java 8 and weren't sure where to land
<enebo[m]>
rdubya: well other than warnings? I have wondered if we get confused by java native exts when switching between 8 and 9+ (thinking one installed with 8 is not there on 9+) but I have run lots of tests with 13
<enebo[m]>
rdubya: all my benchmarking against rails was largely done with 13 so it should be fine
<enebo[m]>
rdubya: but use parallel GC until 14
<rdubya[m]>
cool
<rdubya[m]>
why is 14 better?
<enebo[m]>
work was put into G1 to behave better on startup
<rdubya[m]>
ah ok
<enebo[m]>
we actually would see lower r/s using G1 too so it was not only start up but I don't recall all of what was fixed
<enebo[m]>
feels like I read about that in another century :)
<rdubya[m]>
we're working on optimizing for smaller containers now
<rdubya[m]>
our heap is currently around 1.5 GB I believe, but we want to shrink that
<rdubya[m]>
currently running with Fishwife as our webserver, which is based on Jetty, but looking to swap over to a ruby native webserver
<rdubya[m]>
hopefully that will cut down on the memory usage as well
<enebo[m]>
you guys also limit how long you jit
<enebo[m]>
I have noticed with stuff like redmine actual heap needed is a lot smaller than I expected but that is entirely based on how much live data you work with
<enebo[m]>
Like I could run redmine with a 200M heap
<enebo[m]>
Well I could run with quite a bit less than that but it ran well at 200M
<rdubya[m]>
we're finally getting up to a version where we can test those changes you put in to cap the JIT, its been crazy busy around here
<enebo[m]>
yay in fixing my known problems supporting a[1, &proc{|a| a.stupid_feature }] = 2 spec:ruby:fast hangs and it is not a CPU loop
subbu is now known as subbu|lunch
subbu|lunch is now known as subbu
<lopex>
jeeze just got bitten by 2.7 nil.to_s being frozen
<headius[m]>
No way!
<headius[m]>
How?
<headius[m]>
donv hmmm
<headius[m]>
donv how does before_validation call the block?
<headius[m]>
My guess would be that it passes arguments in a way that doesn't work with a symbol proc
<enebo[m]>
It might be time to boil a tiny sea....
<headius[m]>
If so this might be something that should be fixed in both MRI and jruby
<enebo[m]>
I have always hated block pass being this special nesting
<headius[m]>
Logically I would expect those two lines to work exactly the same
<enebo[m]>
I really feels like AST can just save an iterNode as a peer to normal arguments
ur5us has joined #jruby
<headius[m]>
Oh, I thought you were talking about donv issue
<headius[m]>
Ha
<enebo[m]>
heh no
<enebo[m]>
I have had a similar problem to donv with a suggestion like that but it was for hash which takes two args
<enebo[m]>
that does not look like the issue here though