15:20 <headius> yo yo

15:20 <headius> telegardian: excellent, glad it worked out

15:21 <headius> lopex: they keep sneaking methods from Enumerable into Array

15:30 <headius> enebo: finally looking at struct fix

15:30 <headius> does MRI do something similar for single-arg call?

15:31 <enebo> headius: oh I don't know...I assume so since we do for 1 arg in n-arity version

15:31 <enebo> headius: but I didn't see which commit that logic came from

15:35 <headius> ok

15:35 <headius> well I don't see anything else other tahn Struct needs a rewrite in general

15:37 <headius> oops, I didn't mean to restart that whole build

15:40 <ChrisBr> headius: was playing with benchmkark-driver a little bit, here are some graphs: https://github.com/ChrisBr/benchmark-driver-docker/tree/master/output

15:41 <ChrisBr> for really big hashes, it seems to be twice as fast as 9.2.0.0: https://github.com/ChrisBr/benchmark-driver-docker/blob/master/output/result_100_000_aref.yml-graph.png

16:06 <kares> heh that bm-driver thingy looks need

16:06 <kares> did not know it existed :)

16:09 <headius> ChrisBr: our Struct is some of the oldest code in JRuby and doesn't cache well or integrate with indy well

16:10 <headius> rewrite is perhaps too strong a word

16:10 <headius> those are great numbers btw :-)

16:24 <enebo> https://twitter.com/TruffleRuby/status/1037011265212833792

16:24 <enebo> HJAJAJAJAJAHAAHAHA

16:24 <enebo> I am happy they put out a release but migration guide may be a little ways away

16:27 <headius> well it will be interesting to see if people try to use it

16:27 <enebo> oh yeah I am curious as well. I don't see this being much of an activity until they are running Rails though

16:28 <enebo> We do definitely have some users who are not Rails users but Rails is the elephant in the room

16:28 <enebo> I think it is why it stood out (beyond see our name in someone else release notes

16:36 <xardion> I might give it a stab at some point, I have several non-Rails jRuby projects.

16:38 <xardion> But I'm not itching to jump on that or anything. My Java integration is pretty extensive, and I'd have to refactor basically everywhere to migrate to it

16:40 <xardion> I don't exactly have performance issues with the current jRuby state-of-the art. I can push over 200k messages/sec using 3 VMs.

16:40 <xardion> (and that is reading from Kafka, parsing JSON, performing transforms to create delimited data, and writing it back out to Kafka)

16:48 <xardion> speaking of jRuby and JSON parsing... currently I'm using boon (https://mvnrepository.com/artifact/io.advantageous.boon/boon-json/0.5.2) to do that, and it seems like better options have arisen since I first chose that

16:48 <xardion> was wondering if anyone had used https://github.com/ngs-doo/dsl-json, and if so, opinions?

16:49 <xardion> actually using boon-json 0.6.6, gave an old link

17:05 <ChrisBr> kares: yeah, bm-driver is really neat

17:06 <ChrisBr> I guess it's quite new, heard the first time about it on EuRuKo ...

17:08 <ChrisBr> they also have a github static page where they compare commits and releases: https://benchmark-driver.github.io/

17:11 <kares> najs

17:26 <enebo> xardion: I am porting oj

17:26 <enebo> xardion: but that is obviously ruby json impl

21:26 <headius> https://github.com/jruby/jruby/pull/5307

21:48 <lopex> number

21:48 <lopex> s

21:56 <headius> yeah we have some big compilation opportunities

21:58 <headius> and some really weird perf on GraalEE o_O

22:19 <bomb> hi

22:19 <bomb> is jrubyfx still maintained?

22:25 <headius> bomb: hello again

22:25 <headius> enebo would know better but I think he's checked out for the day

22:25 <headius> if you idle in here or on jruby/jruby gitter you can ask him

22:26 <bomb> headius: hey. sorry i saw your response in #ruby after i disconnected

22:27 <headius> no worries, we have this channel and gitter always on

22:30 <bomb> cool cool. apparently jrubyfx supports javafx 2. but openjfx 11 will be released soon

22:30 <bomb> i think it's best to start writing a thin layer for it

22:35 <headius> oh yeah that would be the next step, need to update jrubyfx to work with openjfx

22:35 <headius> the wrapper layer isn't too thick but I did not do any work on that

22:44 <bomb> cool things are happening in Java world these days

22:45 <bomb> for example, they're working on android/ios: http://openjdk.java.net/projects/mobile/

22:45 <bomb> also new GC is coming, only 1ms pauses! https://www.opsian.com/blog/javas-new-zgc-is-very-exciting/

22:50 <bomb> Java EE is moving to Eclipse Foundation. some servlet containers like WildFly supports JEE 8 now

23:17 <bomb_> oh and of course, Graal :)

23:21 <bomb_> good night!

23:27 <rosenjcb> anyway for me to get a better stacktrace than (NoMethodError) undefined method `[]' for nil:NilClass with JRuby::Rack::ErrorApp::ShowStatus?

23:27 <rosenjcb> I'm working on this project now and I'm getting this error. I have no idea which empty array is being called so I don't exactly know how to debug this issue.]

23:28 <headius> rosenjcb: that's all you get?

23:28 <rosenjcb> Yup, that's all I get

23:29 <rosenjcb> I don't know if this is a jrack thing or if it's just the particular project I'm working on

23:29 <rosenjcb> But I was asking around and my ruby friends say it should at least print out the line where the illegal array call occurs

23:32 <headius> hmmm what's the server?

23:32 <headius> often there's other logs where the full trace might go

23:32 <headius> I assume this is deployed to a Tomcat or something

23:33 <rosenjcb> jetty I believe

23:35 <rosenjcb> What code creates the xhr.responseText var? I'm thinking of trying to just find where that code is generated and then finding an option to get a full stacktrace instead of whatever I have now

23:36 <rosenjcb> Or some kind of log

23:42 <rosenjcb> I'm looking here atm to see if there's anything I can do: https://github.com/tabulapdf/tabula/blob/9849c4f93e7b329e7aad48f770287dbd2cdba8a3/webapp/tabula_web.rb

23:45 <rosenjcb> headius, does that look like a normal stacktrace? I have no idea what to look for to figure out whether it is just an abbreviation. I know that the 500 error is not a predefined error. Looks like a catch all for any runtime exception.

23:46 <rosenjcb> And some users just don't get any error information at all. It just prints out some standard boilerplate with an empty <info> tag for the responseText.