tjohnson has quit [Quit: Connection closed for inactivity]
enebo has joined #jruby
camlow325 has quit []
e_dub has quit [Read error: Connection reset by peer]
e_dub has joined #jruby
enebo has quit [Quit: enebo]
pawnbox has quit [Read error: Connection reset by peer]
<headius>
chrisseaton, subbu: I could see it needing them for debugging but the JVM won't debug undeclared variables either
<headius>
at least the Java-level JVM debuggers can't track variables not in the lvar table for the method, presumably *because* there will be variables introduced by javac
<headius>
I'm also not sure I see why a JVM local would have to be "live"....when you attach a debugger, Hotspot deoptimizes a lot of code anyway
<GitHub17>
[jruby] Jeiwan opened issue #4028: Base64.decode64 incorrect result when missing paddings https://git.io/vKQir
raeoks has joined #jruby
zacts has quit [Quit: WeeChat 1.4]
Uday has joined #jruby
<Uday>
I have sent the following message to jruby mailing list but it is waiting for approval. So posting it here to see if someone can help.
<Uday>
Hi,
<Uday>
We are getting a lot of Java::JavaNioChannels::NotYetConnectedException when doing a http connection. This happens only when there is lot of load. I could reproduce it with production machine configuration and with certain number of concurrent requests in load testing environment.
<Uday>
First few lines of exception stack trace are:
<Uday>
from SocketChannelImpl.java:274:in `ensureWriteOpen'
Uday has quit [Excess Flood]
Uday has joined #jruby
<Uday>
We are getting a lot of Java::JavaNioChannels::NotYetConnectedException when doing a http connection. This happens only when there is lot of load. I could reproduce it with production machine configuration and with certain number of concurrent requests in load testing environment.
yfeldblum has joined #jruby
<Uday>
While analysing, I noticed that RubyTCPSocket creates non blocking socket and although the methods called like connect, select return boolean values, the return value is not checked (code link here). I am suspecting if there could be a scenario where they return false and some other action need to be taken. I am learning more about these apis to de
<Uday>
bug this but can someone with more knowledge about java socket programming confirm if my suspicion is right or not.
<Uday>
o where they return false and some other action need to be taken. I am learning more about these apis to debug this but can someone with more knowledge about java socket programming confirm if my suspicion is right or not.
<Uday>
Even though the error is reproducible on load testing environment, I am not sure how to troubleshoot this issue to get more info. Any suggestion on tools or approach would help.
<enebo>
I do not know how the dating of the jars resolve to the plain SNAPSHOT label but I still have this in my 3.5 million line emacs shell buffer :)
<enebo>
nirvdrum: but the link I gave works so compare this against what you have in .m2 and make sure it is not some other weird crap at play
vtunka has quit [Quit: Leaving]
<nirvdrum>
Thanks. On a call at the moment, but I'll check in a bit.
<GitHub109>
[jruby] joankaradimov opened pull request #4031: Make JRuby compatible with capybara-webkit (and possibly other gems) (master...capybara-webkit-compatibility) https://git.io/vK7pv
<GitHub59>
jruby/master d9203da Subramanya Sastry: Add regression spec for #4014...
<GitHub102>
[jruby] eregon pushed 1 new commit to truffle-head: https://git.io/vK5Um
<GitHub102>
jruby/truffle-head d2f2fcf Benoit Daloze: [Truffle] Do not use the build pack on truffle-head as it won't contain the latest truffle.
<nirvdrum>
enebo: I'd love to know what our build is actually doing. I think chrisseaton reported the crazy amount of stuff being downloaded to mkristian once before. I just blew away my ~/.m2/repository and rebuilt jruby. It fetched 7 different versions of jcodings.
<headius>
zacts: sure, we have lots of folks combining them
<nirvdrum>
13 versions of jffi.
<headius>
usually JRuby is used as the front-end language with Clojure doing some data-processing or other business logic in the backend
<chrisseaton>
nirvdrum: yeah, it's all listed in that build pack repo
<chrisseaton>
nirvdrum: about 400 MB I think
<headius>
nirvdrum: some of that is caused by open-ended dependencies
<chrisseaton>
d
<headius>
but we shouldn't have many (any) of those
<headius>
it may also be negotiating many projects that have different deps
<headius>
like all the JNR projects may not be up-to-date on same jffi
<nirvdrum>
headius: Wouldn't one of them satisfy the requirement though?
<headius>
zacts: all the JVM languages are usable from JRuby without a lot of fuss...it's one of our biggest value-adds
<headius>
nirvdrum: a lower one might satisfy earlier requirements, and then a later one needs a newer jffi
<headius>
but I would have thought that would resolve to one artifact earlier in the build
<headius>
you should post your build log
<headius>
or what's the maven command for printing dependency tree?
<enebo>
yeah it was not long ago we resolved the whole jnr-* chain of deps for a core build
<lopex>
mvn project-info-reports:dependencies gives a full report
pawnbox_ has quit [Remote host closed the connection]
blaxter has quit [Ping timeout: 250 seconds]
qmx has joined #jruby
<nirvdrum>
enebo: It was my fault. I don't think "mvn clean" deletes the jruby-complete.jar and it wasn't being overwritten either for whatever reason.
pawnbox has joined #jruby
thedarkone2 has joined #jruby
mattwildig has joined #jruby
mattwildig has quit [Remote host closed the connection]