<adam12>
headius: That was my first guess - I actually swapped them around to see if that happened.
deobalds has joined #jruby
Guiz has joined #jruby
Guiz has quit [Ping timeout: 268 seconds]
deobalds has quit [Quit: Computer has gone to sleep.]
deobalds has joined #jruby
deobalds has quit [Quit: Computer has gone to sleep.]
deobalds has joined #jruby
Guiz has joined #jruby
Guiz has quit [Ping timeout: 240 seconds]
Guiz has joined #jruby
Guiz has quit [Ping timeout: 248 seconds]
tux_ has joined #jruby
tux_ has quit [Quit: Leaving]
Guiz has joined #jruby
lopex has quit [Quit: Connection closed for inactivity]
Guiz has quit [Ping timeout: 255 seconds]
Guiz has joined #jruby
Guiz has quit [Ping timeout: 255 seconds]
olle has joined #jruby
claudiuinberlin has joined #jruby
olle has quit [Quit: olle]
olle has joined #jruby
eonwe has joined #jruby
<eonwe>
I haven't delved deeploy into Ruby Enumerator#next specs, so can someone point me to a reason why JRuby implemented it through threads (ThreadedNexter)? We've been running into Rails CSRF check issues mentioned in https://github.com/jruby/jruby/issues/2577
drbobbeaty has quit [Read error: Connection reset by peer]
drbobbeaty has joined #jruby
<headius>
adam12: ok, if there's still a behavioral difference we should figure that out
<adam12>
headius: I'll see if I can shorten my example to see if it's a scoping bug or what.
<headius>
eonwe: yes, there's no easy way to support continuations on the JVM right now
<adam12>
Clarifying the format method with Kernel.format fixed it, so either there's another format method that has precedence (the one in Strings.java?) or it's because I was using format as a keyword arg and the scope is off and it was calling itself?
<headius>
the use of threads is unfortunate...we try to pool them so we don't constantly have to spin up new ones, but if you have dozens or hundreds of Enumerato#next going it will hurt
<headius>
adam12: yeah we should figure out what format it was actually calling
<headius>
maybe something that does to_a on the hash and so the order was weird?
<adam12>
headius: Let me know if you want me to post this stacktrace anywhere.
<headius>
adam12: you have a bug open right?
<adam12>
headius: Not yet.
<headius>
ok, open a new bug then, reference this other one and add the full trace as a gist
<enebo>
wow
<adam12>
sigh. This sucks.
<enebo>
adam12: I am also confused about this ... that format one-liner also works for me
<adam12>
enebo: Yeah
<adam12>
Are you using ruby or jruby as command?
<enebo>
jruby
<enebo>
both work for me too though
<adam12>
Well, I'm not sure what's up
<adam12>
but now both are working for me?
<enebo>
oh really?
<adam12>
But weren't a few minutes ago, in same terminal window
<enebo>
adam12: did you reset JRUBY_OPTS?
<enebo>
adam12: or nuke some gems
<adam12>
Nope.
<enebo>
adam12: that is just such a weird conflict
<enebo>
foo() should never conflict with any variable because of the ()
<adam12>
Yeah. I can't reproduce it now.
<adam12>
So maybe related
<adam12>
When I was building bug report data, jruby -v failed with this error
<adam12>
Error opening script file: ruby (No such file or directory)
<adam12>
(so I ran jruby --version without error)
<adam12>
And now, jruby -v runs without fail.
<enebo>
perhaps we use format in -v?
<enebo>
although --version and -v seem like they are aliases
<GitHub167>
[jruby] headius closed issue #4847: YAML parses okay on CRuby, raises exception on JRuby https://git.io/vFgbW
<GitHub22>
[jruby] headius closed issue #2199: Yaml containing escaped single quote fails to parse, works under MRI https://git.io/O0N0hw
<adam12>
It could be related to my env I guess, but I'm not doing anything but chruby-exec in a fresh shell.
<enebo>
adam12: jruby -v by itself should not even startup our runtime
<enebo>
adam12: and it uses String.format (just looked it up)
<enebo>
adam12: if you used -v + -e or somethign then it should have printed out version string and then raised an issue
<adam12>
enebo: That error is probably safe to ignore - after jruby -v failed, I had ran `env` out of curiosity, and missed this little nugget: RUBYOPT=-- ruby -e p JRuby.reference(method(:format)).getMethod.getClass
<adam12>
Why sprintf was randomly failing but works now is still beyond me.
<headius>
heh
<headius>
computers
<adam12>
lol yeah
<headius>
adam12: JRUBY_OPTS=-Xdebug.loadService will show what libraries are getting loaded...maybe we can see if there's some require chruby adds
<enebo>
adam12: so my thought on this now is you had it fail with -v which does not even startup up jruby runtime AND you had it presumably fail in jruby when it was using String.format makes me think it is an ENV issue with Java itself
<enebo>
both would point to weird error from String.format executing
<headius>
adam12: I'd still like to see that full trace with or without an issue
<adam12>
I also see that chruby won't fuzzy-match jruby versions properly
<adam12>
9.1.8 has higher precedence than 9.1.14, so that doesn't help things (when I was just using chruby jruby-9.1)
<headius>
adam12: maybe an issue with alphabetizing the version numbers
<headius>
MRI hasn't had a two-digit version component...ever?
<headius>
they used to have a design flaw that prevented it
<adam12>
I'll keep digging, but maybe safe to assume enebo did fix it and I've just had a shitty day :)
<enebo>
adam12: ok that stacktrace changed my theory but I have no idea how that could happen
<enebo>
adam12: yeah running into this stuff is no fun either
<headius>
gist doesn't help me
<headius>
it sure seems like it's just picking up an older JRuby
<adam12>
headius: It's possible, now that I've seen chruby's matcher.
<adam12>
I could of swore I had the version out put with the error tho
<headius>
yeah seems likely...and probably a bug to report with them if so
shellac has joined #jruby
<headius>
shouldn't be a hard fix for them to sort by numeric components rather
<enebo>
chruby is in ruby isn't it?
<adam12>
shell script
monkstone has joined #jruby
monkstone has left #jruby [#jruby]
<adam12>
chruby could use some love too it seems.
<headius>
adam12: at this point, don't bother with a JRuby issue unless you can reliably reproduce it with JRuby 1.14 or 9.1 branch (outside of a ruby switcher etc)
<adam12>
headius: My tests are passing on latest 9.1 release so I'm happy. We'll chalk it up to computers sucking and move on. Thanks for debugging.
<headius>
ok
<headius>
kares: I just noticed we're up to 8 pending JI specs now
<headius>
I know I probably marked most of these pending but we should tidy that up
<headius>
looks like a lot of type conversion failures, probably things we clarified in newer releases?
<GitHub131>
[jruby] headius pushed 1 new commit to jruby-9.1: https://git.io/vbZWU
<GitHub131>
jruby/jruby-9.1 fab6c5a Charles Oliver Nutter: Properly test this and make it asserts since it should never fail....
<GitHub33>
[jruby] headius closed issue #4880: Dead Exception in RealClassGenerator https://git.io/vbqbX
drbobbeaty has quit [Read error: Connection reset by peer]
drbobbeaty has joined #jruby
<GitHub54>
[jruby] headius closed issue #4879: Performance issue with ruby meta programming https://git.io/vbtAf
drbobbeaty has quit [Read error: Connection reset by peer]
drbobbeaty has joined #jruby
shellac has quit [Quit: Computer has gone to sleep.]
subbu is now known as subbu|lunch
drbobbeaty has quit [Read error: Connection reset by peer]
shellac has joined #jruby
bbrowning is now known as bbrowning_away
<GitHub156>
[jruby] headius closed pull request #4881: Don't unconditionally emit debug level info from RACC (jruby-9.1...jruby-9.1-dont-emit-debug-in-racc) https://git.io/vbm89
<GitHub26>
[jruby] headius pushed 2 new commits to jruby-9.1: https://git.io/vbZup
<GitHub26>
jruby/jruby-9.1 de12ad3 Matthaus Owens: Don't unconditionally emit debug level info from RACC...
<GitHub26>
jruby/jruby-9.1 80b3b3d Charles Oliver Nutter: Merge pull request #4881 from haus/jruby-9.1-dont-emit-debug-in-racc...
subbu|lunch is now known as subbu
shellac has quit [Ping timeout: 252 seconds]
drbobbeaty has joined #jruby
jmalves has joined #jruby
drbobbeaty has quit [Read error: Connection reset by peer]
<headius>
lopex: test_regsub_K passes on master o_O
<lopex>
headius: yeah, small fix to gsub
<headius>
oh you fixed on 9.1 too?
<lopex>
on master
<headius>
you can cherry-pick to 9.1 if it's not a big change
<headius>
I just noticed it passing while I was doing 9.1 merge
<lopex>
headius: same fix mri did actually
<headius>
haha
<headius>
ok I'll handle it
<lopex>
pick two commits, the other was untag
<lopex>
headius: is there any automatic untagging by default ?
nlew has joined #jruby
<GitHub175>
[jruby] headius pushed 2 new commits to jruby-9.1: https://git.io/vbZy4
<GitHub175>
jruby/jruby-9.1 07f9e65 Marcin Mielzynski: fix test_regsub_K
<GitHub175>
jruby/jruby-9.1 bede47a Marcin Mielzynski: untag test_regsub_K
<GitHub79>
[jruby] headius pushed 16 new commits to master: https://git.io/vbZyB
<GitHub79>
jruby/master b01a8c4 Charles Oliver Nutter: Use superinterface here to catch all jnr-enxio channels....
<GitHub79>
jruby/master ab8e774 Charles Oliver Nutter: Generic a few parts of RubyArray.
<GitHub79>
jruby/master e63afd3 Charles Oliver Nutter: Formatting.
<headius>
lopex: no, because some are intermittent or fail only in specific envs
<headius>
enebo: hey I just noticed we have ObjectSpace.garbage_collect set to no-op too
<enebo>
hmm
<headius>
I searched through many pages of "ObjectSpace.garbage_collect" on github and didn't see many real Ruby uses, so I think we should change this to actually call GC
<enebo>
yeah determinism cannot really be counted on there but doing something maybe is fine like a gc :)
<headius>
this may be one reason a couple specs are flaky...MRI uses ObjectSpace.garbage_collect rather than GC.start...we patch the latter but not the former
<enebo>
ah
<headius>
GC.start is used a lot in the wild but this other one is not
<headius>
either way people shouldn't be calling GC but I feel like more people doing ObjectSpace.garbage_collect really mean it than those who have GC.start
ralphw has joined #jruby
<ralphw>
Hi
<ralphw>
Are there any known issues installing JRuby via RVM on Windows 10 using the linux sub system?
<headius>
ralphw: not to my knowledge
<GitHub18>
[jruby] headius pushed 1 new commit to jruby-9.1: https://git.io/vbZHd
<GitHub18>
jruby/jruby-9.1 a25301e Charles Oliver Nutter: Tag spec that depends on GC behavior and fails intermittently.
<headius>
I'm not sure how well Java works in the linux subsystem but that's separate
<enebo>
ralphw: I guess I got 9.1.8.0 but it ran enough to print out a string
<enebo>
ralphw: it also means native probably loaded since I think we do not run well in 9.1.x if our native backend fails to load
<GitHub59>
[jruby] headius pushed 1 new commit to jruby-9.1: https://git.io/vbZA7
<GitHub59>
jruby/jruby-9.1 307f557 Charles Oliver Nutter: Tag off a couple other specs that depend on GC behavior.
<GitHub44>
[jruby] headius pushed 1 new commit to jruby-9.1: https://git.io/vbZx9
<GitHub44>
jruby/jruby-9.1 0e6efdd Charles Oliver Nutter: Re-add an mbean for monitoring cache invalidations....
drbobbeaty has quit [Read error: Connection reset by peer]
<headius>
enebo: nice
<headius>
after 9.1.15 maybe I'll have time to work on Windows support
<headius>
although this is really Linux we're talking about
<GitHub83>
[jruby] headius pushed 1 new commit to jruby-9.1: https://git.io/vbZp9
<GitHub83>
jruby/jruby-9.1 a8bfd60 Christian Meier: pick new jruby-readline...
<GitHub184>
[jruby] headius closed issue #4740: activation of jar-dependencies in bundler clashes with shipped gem https://git.io/v7NT2
<GitHub147>
[jruby] headius pushed 1 new commit to jruby-9.1: https://git.io/vbnJC
<GitHub147>
jruby/jruby-9.1 fd1f47a Charles Oliver Nutter: Avoid deadlocks in IO by only locking one thing. Fixes #4854...
<GitHub119>
[jruby] headius closed issue #4854: Simultaneous client reads / writes to the same socket from different threads can deadlock https://git.io/vFKz5
<GitHub83>
[jruby] headius closed issue #4856: Opaque `SystemCallError` in Conjunction with `Puma::ConnectionError: Socket timeout writing data` https://git.io/vFyf2