yfeldblum has quit [Remote host closed the connection]
camlow325 has quit [Remote host closed the connection]
rsim has joined #jruby
camlow325 has joined #jruby
pitr-ch has quit [Ping timeout: 272 seconds]
camlow325 has quit [Remote host closed the connection]
pitr-ch has joined #jruby
skade has joined #jruby
yfeldblum has joined #jruby
deobalds has quit [Quit: Computer has gone to sleep.]
yfeldblum has quit [Read error: Connection reset by peer]
yfeldblu_ has joined #jruby
yfeldblu_ has quit [Read error: Connection reset by peer]
yfeldblum has joined #jruby
marr has joined #jruby
shellac has joined #jruby
benlovell has joined #jruby
drbobbeaty has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] mkristian closed issue #2873: Class not found exception on ppc64le http://git.io/vf2NU
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] mkristian reopened issue #2873: Class not found exception on ppc64le http://git.io/vf2NU
JRubyGithub has left #jruby [#jruby]
KevinCorcoran has joined #jruby
KevinCorcoran has quit [Ping timeout: 240 seconds]
deobalds has joined #jruby
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
shellac has quit [Quit: Computer has gone to sleep.]
benlovell has quit [Ping timeout: 255 seconds]
benlovell has joined #jruby
shellac has joined #jruby
pjammer has joined #jruby
drbobbeaty has joined #jruby
shellac has quit [Quit: Computer has gone to sleep.]
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
mike_c_11 has joined #jruby
dinfuehr_ has joined #jruby
dinfuehr has quit [Ping timeout: 246 seconds]
dinfuehr_ has quit [Ping timeout: 246 seconds]
benlovell has quit [Ping timeout: 250 seconds]
shellac has joined #jruby
yfeldblum has quit [Ping timeout: 265 seconds]
benlovell has joined #jruby
benlovell has quit [Ping timeout: 256 seconds]
bjfish2 has joined #jruby
skade has quit [Quit: Computer has gone to sleep.]
cremes has quit [Remote host closed the connection]
deobalds has quit [Quit: Computer has gone to sleep.]
nirvdrum has joined #jruby
cremes has joined #jruby
<headius>
good morning
codefinger has joined #jruby
benlovell has joined #jruby
tcrawley-away is now known as tcrawley
skade has joined #jruby
KevinCorcoran has joined #jruby
lance|afk is now known as lanceball
enebo has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum opened issue #2917: [Truffle] NPE in Binding#local_variable_get http://git.io/vUTDD
JRubyGithub has left #jruby [#jruby]
KevinCorcoran has quit [Ping timeout: 272 seconds]
cremes has quit [Quit: cremes]
mje113__ has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] enebo pushed 1 new commit to master: http://git.io/vUTxI
<JRubyGithub>
jruby/master 0ec6cdd Thomas E. Enebo: Use jcodings rather than Java for alnum ident (will work with all encodings without needing charset)
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] bjfish pushed 1 new commit to master: http://git.io/vUTxN
<enebo>
headius: yeah I am looking at this a little but I am not sure I will make progress without epic MRI trawl
donV has joined #jruby
camlow325 has joined #jruby
<headius>
I did not feel like trawling
<enebo>
my gem list is surprisingly small too…I am guessing this just continues to be a drag as things increase…OTOH loadservice is not even in hprof sites so at this point it is may a little over 1%
<headius>
at least, not after I managed to fix the issue with yet another one-off hack
<enebo>
I was just surprised to see a Ruby method in this list at all in gem list
<headius>
this logic is copied from buildEnsure with some of the funk removed
<headius>
we still have $!-resetting happening along other paths, and those could probably go away
<headius>
this is the right approach though, eh? implicit ensure around any rescued region to reset $!
<headius>
I do love that these fixes require no changes in JIT
<headius>
the down side of this is that there's another try/catch around the whole begin+rescue+ensure, but I don't see another way since we weren't setting up individual try/finally to clear $!
<enebo>
headius: originally they were all removed thinking we were doing these resets
<headius>
did it originally do this in IR? I'm pretty sure it did in AST
<enebo>
headius: I don’t really understand the undepinning of the IRBuilder ensure impl that much past cloning
<headius>
the cloning is a minor detail...I'm not doing that because it's just as easy to add new instrs
<headius>
specifically for the non-exceptional reset of $!
<headius>
hmm
<headius>
though maybe I shouldn't have that
<headius>
whatever $! is at the end of the begin region, it should just stay that way
<JRubyGithub>
[jruby] headius created ensure_last_error_reset (+1 new commit): http://git.io/vUIeb
JRubyGithub has left #jruby [#jruby]
<JRubyGithub>
jruby/ensure_last_error_reset 2880c73 Charles Oliver Nutter: Add another layer of try/finally around rescue/ensure to reset $!....
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius opened pull request #2919: Add another layer of try/finally around rescue/ensure to reset $!. (master...ensure_last_error_reset) http://git.io/vUIvv
JRubyGithub has left #jruby [#jruby]
<headius>
nirvdrum: are you fixing that one?
<nirvdrum>
Not at the moment. I'm knee-deep in fixing Proc#parameters with kwargs for Truffle.
<headius>
ok
<nirvdrum>
But I can take a look after.
<headius>
nah, I'm waiting on review from subbu and enebo so I'll hit it
<nirvdrum>
I was trying to see how well JRuby handles kwargs and noticed that mismatch.
<headius>
yeah, enebo got this 100% but probably missed something in JIT
jph98 has quit [Quit: jph98]
camlow325 has quit [Remote host closed the connection]
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/vUIOC
<JRubyGithub>
jruby/master e60d0dd Kevin Menard: Added keyword argument specs for Proc#parameters.
JRubyGithub has left #jruby [#jruby]
<subbu>
I think it works. I was staring at the code y'day night to figure out how to do this so that we don't have to add a generic global wrapper in all cases OR do this for every rescue body.
vikaton has quit []
chamila has joined #jruby
<subbu>
one solution is: this wrapper is only required if the rescue body code has a non-local return (which can be figured out by ast-inspection easily -- probably can be a helper method on the ast).
<subbu>
so, that should probably make it possible to only add the wrapper around rescue bodies that really require it.
<headius>
subbu: well I'm wondering about other cases we can't detect
<headius>
a rescue body could call a method that does a non-local return unrolling all the way out
<headius>
def foo; bar { return }; end; def bar(&b) raise; rescue; baz(&b); end; def baz; yield; end
skade has quit [Quit: Computer has gone to sleep.]
<headius>
there doesn't seem to be any assurance that $! is cleared on exceptional branch out of a rescue
<headius>
only normal exit
<subbu>
hmm .. yes.
<headius>
fact is any call could trigger a stack unroll that's not a Ruby exception
<subbu>
so, we could simplify other IR code if we are going to do this in a global wrapper.
<subbu>
we are adding a global ensure block for other scenarios, and we could piggyback this onto that.
cremes has joined #jruby
<subbu>
but, that can be done in a followup. this $! business is a bit crappy.
<headius>
hmm, wait, mine only works within rescue bodies... compiling $! to look at the local var holding the exception
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/vUICe
<JRubyGithub>
jruby/master f637869 Kevin Menard: [Truffle] Exclude the Proc#parameters tests from running.
JRubyGithub has left #jruby [#jruby]
<headius>
subbu: do you foresee any problems with making a GEB-like wrapper around rescues?
<subbu>
I am in the Bay Area this week .. heading out in another 30 odd mins and will be mostly away from computer till monday.
<headius>
when it's a top-level rescue (full method body) the existing GEB could be used
<subbu>
s/monday/tuesday morning/
<headius>
ok
<subbu>
why not use a top-level rescue? so, that was the second solution i was considering y'day night.
<headius>
def foo; begin raise 1; rescue; end; begin; raise 2; rescue; end; end
<headius>
you can have many rescues in a method body and they each need to reset $! on exit
<subbu>
i mean: a top-level ensure block always .. because no matter what non-local return path requires $! to be reset it can be done once on the way out for all paths?
<headius>
ahh yes
<headius>
that's basically what this does
<subbu>
right.
<headius>
well, top-level meaning wrapped around the begin/rescue/ensure from the AST
<headius>
hmm
<headius>
actually you may have something
<chrisseaton>
Fix for the Travis failure pushed
<headius>
user-defined ensure body doesn't need to change $!
<headius>
chrisseaton: thanks
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] chrisseaton pushed 1 new commit to master: http://git.io/vUIWd
<JRubyGithub>
jruby/master 7dd17d2 Chris Seaton: [Truffle] Better guard for $_ in Binding.
JRubyGithub has left #jruby [#jruby]
<subbu>
so, you can use the existing GEB (or add it always if it is ont being added) to avoid adding multiple onion layers.
<headius>
hmm
<headius>
I don't think that will work because between sibling rescues in a method body, the old $! needs to be visible
<headius>
GEB is too late
<subbu>
but, we are only talking of non-local return paths that will escape the scope.
<headius>
that's true
<subbu>
=> there is no way you can go back to executing other code in that same scope
<headius>
actually you can
<headius>
another ensure
<subbu>
ah, right and it can inspect $!
<headius>
begin; begin; raise; rescue; 1.times { return }; ensure; p $!; end
<subbu>
ugh.
<headius>
er, another end in there
<headius>
but you get it
<subbu>
yes.
<headius>
ugh, I think I just realized my patch is wrong still anyway
<headius>
because it will clear $! for ALL exceptions including the ones that happen in the begin
<nirvdrum>
headius: It's a good thing you fielded that one. My guess is whatever I came up with wasn't going to be nearly as concise.
<headius>
when a higher up rescue looks for $! it won't see it
<headius>
nirvdrum: lucky one
<headius>
subbu: I'll keep working at it
<headius>
but I don't see a way to do it without adding an ensure to every begin/rescue
<subbu>
headius, ok .. if i have any ideas ... i'll post them later, but i imagine you'll figure this out before then.
<headius>
ok
shellac has quit [Ping timeout: 250 seconds]
<subbu>
headius, yes, that is what i kept stumbling into y'day night and i was trying hard not to have to do it.
<subbu>
but, all those other ideas seem broken.
<headius>
down sides to this approach?
<headius>
it's basically what we had in the AST interp
<subbu>
nothing really .. just seems like excess code.
<headius>
if there were a way to access the in-flight JVM exception at any time, we could use that
<mjc_>
huh, getting "java.lang.ArrayIndexOutOfBoundsException: 0" while loading a rails app
<mjc_>
seems like one should not be able to trigger an empty array exception from ruby, heh
<mjc_>
looks like the truffle stuff might also use isAbsolutePath without a length check
<headius>
probably just duplicated from jruby proper
<mjc_>
hmm jruby-openssl won't build for me
<mjc_>
or something?
<headius>
won't build?
<mjc_>
[ERROR] Failed to execute goal io.tesla.polyglot:tesla-polyglot-maven-plugin:0.1.1:execute (install_gems) on project jruby-lib: (InstallError) invalid gem: package is corrupt, exception while verifying: undefined method `path2class' for #<Psych::ClassLoader:0x45f3540> (NoMethodError) in
<headius>
the log you pasted appears to have jruby-head lines in it
<mjc_>
just the folder name
pjammer has joined #jruby
<mjc_>
[INFO] Building JRuby Core 1.7.20
<headius>
yeah the path confused me
<mjc_>
did git checkout 1.7.20 before trying to build
<headius>
let me try here
<chrisseaton>
headius: I'm doing a talk at the Truffle workshop at ECOOP on why we can't re-use more of JRuby in Truffle - shame you won't be there but I'll send you slides of time http://2015.ecoop.org/track/Truffle-2015-papers
dinfuehr has quit [Remote host closed the connection]
dinfuehr has joined #jruby
<nirvdrum>
chrisseaton: Are the sessions being streamed?
<chrisseaton>
no - these academic conferences are almost never filmed let alone streamed
<chrisseaton>
Maybe we can stick a phone on a stick though and record it
<mjc_>
ustream on a phone is a thing
dinfuehr has quit [Remote host closed the connection]
<mjc_>
recording it would probably be much higher quality though
dinfuehr has joined #jruby
<headius>
hmm
<headius>
it build ok here
<headius>
chrisseaton: and what's the tldr on that?
<mjc_>
I blew away ~/.gems/jruby/1.9.3 as well as ~/.m2 and still couldn't build it
<mjc_>
:(
<nirvdrum>
Short talks, eh?
<headius>
mjc_: hmmm
<headius>
mjc_: make sure nothing's odd in your ENV
<chrisseaton>
We've found that Truffle does best when you write the runtime using Truffle, not just the execution part, and the JRuby runtime wasn't designed to work like a Truffle program
<headius>
that error means that psych is not booting completely, but it should be running from a known good jruby release
<chrisseaton>
But I'm going to advocate anyone else trying what we did (Jython) do it more incrementally - you could do just the execution part to start with, and still get good performance
<headius>
chrisseaton: that seems like a flaw in truffle
<headius>
why is truffle-marked-up java code easier to optimize than just java code? Because truffle/graal can't inline normal Java code?
<chrisseaton>
Well you could also describe it as a flaw in C2 or even Graal - if it could type profile, condition profile, type specialise generic data structures etc it wouldn't be needed - but it can't - so we do it using Truffle
<mjc_>
just took a look at the output of set, didn't see anything unusual
<headius>
chrisseaton: ok, that's what I figured
<mjc_>
java 1.8.0_40
<headius>
so because truffle lang and normal java go through different compilers, they can't optimize together
dinfuehr has quit [Ping timeout: 256 seconds]
<headius>
I still see that as a limitation in truffle
<headius>
it basically means you have to do a hard call any time you have to call from truffle lang into java
<headius>
right?
<headius>
it's not a limitation in C2 because indy-based JVM lang can inline and optimize across that boundary
<chrisseaton>
no we can optimise across that boundary fine - but when we do that, the code on the other side - the JRuby runtime code - is more generic than we'd like
<mjc_>
lunch for me while jdk update downloads, to see if that helps, since I'm grasping at straws
<headius>
so existing jruby java code will inline into jruby+truffle code?
<chrisseaton>
Yeah - and we do that in quite a few places
<headius>
ok, then it's not as bad as I thought
<headius>
I guess this begs the question though...why can't truffle see the specializations it needs in jruby's java code
<headius>
seems like a huge limitation if truffle optimizations can only happen when you annotated the java code to death
<chrisseaton>
It's not that - it's that we optimised the control flow stuff and things like local variables - so that the runtime became our next bottleneck
<mjc_>
headius: trying a build on a linux box just to get somewhere
<headius>
chrisseaton: sure, but you can only trufflize so much java
<chrisseaton>
So we needed to do things like specialise data structures, which JRuby hasn't done yet, so we needed to write that ourselves
<headius>
in a system that's 10% ruby and 90% arbitrary third-party Java code, is that going to hurt optimization in the ruby?
<headius>
specialized data structures I get...neither runtime does that automatically right now
<chrisseaton>
You mean like running external Java libraries? Performance won't be any worse than JRuby in that case
<headius>
neither jvm-level runtime I mean
<mjc_>
oh yeah, same array index thing happens inside a centos 6.5 vagrant box
lanceball is now known as lance|afk
<headius>
chrisseaton: so is the real bottom line that you needed more specialized data structures to plug into truffle-specialized code?
<headius>
or at perhaps data structures that had the hooks needed for truffle to specialize them into the lang's code
<chrisseaton>
headius: well part of what I'm saying is that it would be have been fine to use the JRuby runtime first, but when we merged we were already beyond that point - but then to add optimisations like our hidden-class object it's hard to integrate that into existing code
marr has quit [Ping timeout: 240 seconds]
<headius>
ok
<headius>
then perhaps the paper shouldn't be why you "can't" use more of JRuby core
<headius>
you could
<headius>
I was confused by the word "can't"
<headius>
the real title to me is "why unspecialized data structures don't allow us to reach full performance in a truffle-based language"
<headius>
and that's not really different from any language, then
<headius>
eh?
<chrisseaton>
The actual title is just 'Integrating Truffle into an Existing Language Implementation' - we can't for where we currently are in this project, as we've moved so far beyond that point and we're aiming to optimise things that break compatibility with your runtime - like using our own object model
benlovell has joined #jruby
<headius>
chrisseaton: ok, so it's as much about logistical reasons as anything
<headius>
how much of the data structure specialization depends on truffle or graal?
<headius>
i.e. is any portion of it usable on a non-graal VM?
<chrisseaton>
the object model probably is - but I haven't had time to integrate that into JRuby yet
<headius>
that would be very interesting
<headius>
you've seen my prototype of specializing ivars I assume
<chrisseaton>
yeah
<headius>
it definitely works...but I have never had time to push forward
<chrisseaton>
it's hard for me to justify working on that for a long period of time - but if you had a volunteer we could give them all the support they need to do it
<headius>
depends how isolated it is from truffle-specifics
<headius>
mjc_: ok, back to you
benlovell has quit [Ping timeout: 246 seconds]
<headius>
chrisseaton: I'll stop nitpicking now and look forward to the slides
<headius>
obviously anything that can be used well without graal VM we'd be interested in using in JRuby proper
<mjc_>
that sounds like a neat project and I hope I could find the time to learn enough java to volunteer :P
<headius>
mjc_: maven version?
bbrowning is now known as bbrowning_away
<mjc_>
headius: 3.3.3
<headius>
wow, even newer than mine
colinsurprenant has joined #jruby
<headius>
mjc_: can you make sure there's no psych gem installed in your gems for that env? I think that's the problem
<headius>
you switched ruby-head to 1.7 but there's psych gem from 9k build
<headius>
psych gem does not work on 1.7
<headius>
and this is the error you'd get, I think
<headius>
plus it breaks yaml completely because it doesn't boot, so RubyGems breaks and reports weird errors
<mjc_>
I blew away $GEM_HOME and got it
<headius>
mjc_: can you try to build a clean clone with a clean env, just to humor me?
<mjc_>
oh weird, I see
<headius>
it really looks like it's trying to use psych gem on 1.7
<mjc_>
there was stuff in lib/ruby/gems in my repo
<mjc_>
git clean -dfx seems to have fixed it
<mjc_>
so far
<mjc_>
no, nevermind
<headius>
I was just goign to suggest that
<mjc_>
wiping it out and trying again
<headius>
I also realized psych gem is installed into stdlib and switching branch wouldn't clear that
<headius>
so a clean -dxf with no gems anywhere in env still fails to build 1.7.20?
<mjc_>
I unset GEM_* variables and did git clean -dfx; mvn clean -Pbootstrap; mvn and it failed
<mjc_>
just deleted the dir and am re-cloning, will take a while
<headius>
I'm trying -Pbootstrap now, did not before
<mjc_>
oh and blew away ~/.m2
<headius>
ok
<headius>
if that fails for you I'll be really confused
<headius>
unless this fails for me
<headius>
yeah worked ok
<headius>
mjc_: make sure there's no other "ruby" or "jruby" in patht oo
<headius>
too
<mjc_>
I used chruby to put this folder at the front of the path
<mjc_>
but I tried it without chruby loaded
<mjc_>
ok finally checked out again
<headius>
ok
benlovell has joined #jruby
<headius>
I wiped out .m2/repository, clean -dxf, mvn clean package -Pbootstrap, and it worked ok again...starting to wonder if it's mvn version
<headius>
still worked ok
<headius>
oops
djbkd has quit [Remote host closed the connection]
<headius>
I will try updating mvn
<mjc_>
it's doing the redownload-all-the-deps thing now
<mjc_>
I'm less concerned about not being able to build 1.7 than about this other stupid bug heh
<headius>
yeah I understand
<headius>
we're making a nice yak coat today
<mjc_>
hahah yes
<headius>
I kinda want to jump on build problems like this, though, because they keep coming back
<mjc_>
ok it built
<mjc_>
fresh repo did the trick
<mjc_>
so something that was living in my checkout refused to die even with git clean -dfx and mvn clean -Pbootstrap, huh
djbkd has joined #jruby
<headius>
mjc_: yeah weird
<headius>
I will make a bug to tag the psych gem as requiring 9k
<headius>
if that's possible
<mjc_>
oh, wow
<mjc_>
SIGBUS from freshly built 1.7.20
<mjc_>
oh id10t error
<mjc_>
was running maven
<headius>
ahh ok :-)
<headius>
you're raising my blood pressure today
<mjc_>
sorry :'(
<headius>
it's ok...I'm glad we still just have the one bug :-)
benlovell has quit [Ping timeout: 250 seconds]
<mjc_>
ok glob_helper gets called twice
<mjc_>
I don't understand why it's blowing up
<mjc_>
I guess the println isn't instructive enough
<mjc_>
ok so for some reason bytes ends up 0 length, as suspected
<mjc_>
ah, something is passing 0 for end
mistergibson has quit [Quit: Leaving]
<headius>
mjc_: I don't suppose you've ever set up an IDE to step-debug jruby, have you?
<mjc_>
so I guess I can put platform: :ruby on spring for now and that will solve my issue in the short term
<mjc_>
but that still looks like a jruby bug, heh... I don't know how to get what you're asking for and now that I have a workaround I don't think I can help find it further until after work hours :(
<headius>
mjc_: so it's something in spring triggering it eh?
travis-ci has joined #jruby
<mjc_>
it's a combination of spring and dotenv-rails
<mjc_>
headius: my suspicion is that spring messes with the way things are loaded and dotenv messes with the ENV, so between the two something gets fubared
<headius>
that's only failure on master travis right now
<enebo>
I did not test this but I though this would accept more output
<enebo>
err input
<headius>
you might be interested in some of the argument refactoring I just did too, trying to clean up the various ways we reprsent Method#parameters
dinfuehr has joined #jruby
<enebo>
headius: sure btw I did change some stuff last time to pass tests but I mostly fixed arity
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius pushed 1 new commit to master: http://git.io/vUt8x
<JRubyGithub>
jruby/master 60b462e Charles Oliver Nutter: Refactor new Argument stuff and clean up some related logic....
JRubyGithub has left #jruby [#jruby]
<headius>
we had different formats for IRMethodArgs.getParameterList and MethodArgs2.getParameterList
<enebo>
headius: parameters themselves I did fix some issues but the order is different than MRI for example so I know we had some issues
<headius>
I made the IR version actually be a structured representation and changed the name
<headius>
then tidied up the logic converting one form to another
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/vUt4x
<JRubyGithub>
jruby/master 1493648 Kevin Menard: [Truffle] Implemented Method#parameters and UnboundMethod#parameters.
JRubyGithub has left #jruby [#jruby]
<enebo>
headius: do we still have multiple types?
<headius>
enebo: yes :-(
<enebo>
err impls
<enebo>
yeah that could change
<headius>
some code uses the prefixed form
<headius>
I did not have code to convert from prefixed form to Argument form, so I didn't elimninate that
<headius>
but it would be a good change
<headius>
I was going to fix the parser bug first
<nirvdrum>
My curse is your gain.
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] enebo pushed 1 new commit to master: http://git.io/vUtBb
<JRubyGithub>
jruby/master 48120ba Thomas E. Enebo: Revert "Use jcodings rather than Java for alnum ident (will work with all encodings without needing charset)"...
JRubyGithub has left #jruby [#jruby]
<enebo>
HEH
<enebo>
I fixed it real good
<headius>
was that a fix for a bug, or just cleanup?
<enebo>
headius: It was realization that if source is weird iso_8859_xx it may not recognize 7bit identifiers if JVM does not have charset support
<enebo>
headius: what I changed it to is what MRI actually has
<headius>
ahh sure
<headius>
which 8859 was it?
<enebo>
I don’t know
<headius>
I implemented one charset that Java was missing
<enebo>
but I imsagine there may be one
<headius>
dunno if that's enough
<headius>
15 I think
<enebo>
any 7bit which has more than a-z
<enebo>
at least I was just thinking about it without truly knowing for sure
<enebo>
perhaps all 7bit whch has ascii will never have other chars and go mbc for any others
<enebo>
I suspect there is some encoding though
<enebo>
fwiw, if you are in FooBaristan and have extra letters no doubt your JVM also does
<enebo>
so it might not be so important
<enebo>
Plus we have issue that we intern names as Java strings
<enebo>
but it still felt more correct
skade has joined #jruby
<mjc_>
headius: do you want me to make a bug report for the dotenv/spring thing?
<headius>
mjc_: yeah please
<headius>
thank you for getting a repro
<headius>
enebo: ok, I understand
<mjc_>
no problem, sorry I wasn't able to track it down further, I would have rathered to toss you a patch :(
<headius>
s'ok, glob is pretty nasty
djbkd has quit [Remote host closed the connection]
<nirvdrum>
headius: They're out to get us. I have jury duty in two weeks. I opted not to postpone because of upcoming travel.
<headius>
that's weird
<headius>
lucky us
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] mjc opened issue #2922: JRuby 1.7.x crashes with java.lang.ArrayIndexOutOfBoundsException: 0 when booting rails with spring and dotenv-rails loaded http://git.io/vUtad
JRubyGithub has left #jruby [#jruby]
<headius>
yeah if I weren't going to be in Estonia that week I might not have postponed, but then after that I've booked up too
<headius>
mjc_: thanks!
<mjc_>
anytime
yfeldblum has quit [Remote host closed the connection]