<nirvdrum>
headius: Awesome. I tried tracing through, but I couldn't keep track of it all. And I tried on different boundaries, but couldn't always induce it.
<nirvdrum>
I'm happy to just find and report the issues if you're cool with fixing them, though :-)
drbobbeaty has quit [Read error: Connection reset by peer]
kfpratt has quit [Ping timeout: 272 seconds]
skade has quit [Quit: Computer has gone to sleep.]
mcclurmc has joined #jruby
mcclurmc has quit [Ping timeout: 252 seconds]
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
nirvdrum has joined #jruby
unarmed has joined #jruby
universa1 has left #jruby ["Verlassend"]
pitr-ch has joined #jruby
PragTob has quit [Remote host closed the connection]
erikhatcher has quit [Quit: erikhatcher]
<unarmed>
hi, i've read that c-extensions are not really supported by jruby in java anymore. any way to get this working for my project? I'm embedding the jekyll gem inside a maven plugin
iamjarvo has joined #jruby
erikhatcher has joined #jruby
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
mitchellhenke has quit [Quit: Computer has gone to sleep.]
mitchellhenke has joined #jruby
bjfish2 has joined #jruby
kfpratt has quit [Remote host closed the connection]
rsim1 has quit [Read error: Connection reset by peer]
kfpratt has joined #jruby
<bjfish2>
Should this be Bignum? “puts (-2**63).class”
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum created truffle_taint_exemptions (+1 new commit): http://git.io/j6X0
<JRubyGithub>
jruby/truffle_taint_exemptions bd06055 Kevin Menard: [Truffle] Don't taint from self in String#[](String).
JRubyGithub has left #jruby [#jruby]
<chrisseaton>
bjfish2: that's Long.MAX_VALUE (in case you weren't aware) - so I'd say no - maybe it's to match MRI?
nirvdrum has joined #jruby
<chrisseaton>
bjfish2: oh it's Fixnum in JRuby - that's what I'd expect then
iamjarvo has joined #jruby
<chrisseaton>
bjfish2: the reason it is Bignum in MRI is due to tagging their integers are less than 64bit
<bjfish2>
chrisseaton i think long max value is 2^63-1, so it’s one more than max value
<bjfish2>
chrisseaton sorry it is min value -2^63
<chrisseaton>
yeah
<chrisseaton>
I meant Long.MIN_VALUE, sorry
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum force-pushed truffle_taint_exemptions from bd06055 to 3ec3437: http://git.io/j6Ml
<JRubyGithub>
jruby/truffle_taint_exemptions 3ec3437 Kevin Menard: [Truffle] Don't taint from self in String#[](String).
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum opened pull request #2780: [Truffle] Don't taint from self in String#[](String). (master...truffle_taint_exemptions) http://git.io/j6Mg
JRubyGithub has left #jruby [#jruby]
<bjfish2>
chrisseaton my real issue is: puts (-2**63).abs is -9223372036854775808
havenwood has joined #jruby
shellac has joined #jruby
<chrisseaton>
bjfish2: in Truffle? yeah that's weird - it's a bug
<bjfish2>
chrisseaton truffle only yeah i’ll log an issue, i’m not seeing what the issue is
<chrisseaton>
bjfish2: looks like we're not checking for overflow in FixnumNodes.AbsNode
marr has joined #jruby
<chrisseaton>
do you want to either try to fix or open an issue?
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] bjfish opened issue #2781: [Truffle] Expected Absolute Value for Fix http://git.io/j6yc
JRubyGithub has left #jruby [#jruby]
mkristian has joined #jruby
<bjfish2>
chrisseaton i took a look, not really seeing whats wrong, so I’ll open an issue, this is Bignum in MRI BTW, looks like Fixnum in jruby
<chrisseaton>
What is a Fixnum and what is a Bignum is a property of the architecture - so I think we're free to set our own limits on the Java 'architecutre'
<bjfish2>
chrisseaton okay, I understand now, also i understand too how calling abs on -2**63 would overflow long
Hobogrammer has joined #jruby
mkristian has quit [Ping timeout: 245 seconds]
gregorsc5 has quit [Quit: Leaving.]
triple_b has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<lopex>
chrisseaton: I guess you'll put as much effort specializing complex/rational as you do for fixnum ?
tcrawley is now known as tcrawley-away
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/jPHA
<JRubyGithub>
jruby/master fe5b4b4 Kevin Menard: [Truffle] Pulled in almost all of Rubinius's String....
JRubyGithub has left #jruby [#jruby]
<nirvdrum>
lopex: Currently we're doing Rational all in Ruby, I believe.
tcrawley-away is now known as tcrawley
<nirvdrum>
But if we need to move some of that to Java, we will.
<lopex>
nirvdrum: ah, sounds reasonable
<nirvdrum>
lopex: At this point, if something isn't stressed in bench9000, we're not putting much emphasis on making it fast.
<lopex>
yeah
<lopex>
always real bottlenecks
<lopex>
*real world
<chrisseaton>
Fixnum has a machine representation as well - so we want to use that where possible - for complex and rational there isn't one so there's nothing to try to get to in the same way
tcrawley is now known as tcrawley-away
tcrawley-away is now known as tcrawley
kfpratt has joined #jruby
kfpratt has quit [Remote host closed the connection]
kfpratt has joined #jruby
mattwildig has joined #jruby
<nirvdrum>
lopex: Why doesn't String#each_byte need to take the bytelist begin value into account?
fivebats has joined #jruby
mcclurmc has quit [Remote host closed the connection]
<lopex>
nirvdrum: enumerateBytes ?
<nirvdrum>
Yeah.
<nirvdrum>
It always starts at 0.
<lopex>
it was rewritten by Charlie at some point later on
<lopex>
but I'd say it should obey begin
<lopex>
oh it depends if get is begin aware
<lopex>
is it ?
<lopex>
yeah, it is
<lopex>
nirvdrum: ^^
<nirvdrum>
Ahh, I missed that. Sorry.
tharindu has quit [Ping timeout: 252 seconds]
calavera has joined #jruby
<nirvdrum>
lopex: Next silly question . . . why return an int there instead a byte?
<lopex>
nirvdrum: from where ?
kares has joined #jruby
<nirvdrum>
ByteList#get
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] derek-upham opened issue #2783: jruby.bash includes current directory in classpath, should not http://git.io/jXvE
JRubyGithub has left #jruby [#jruby]
iamjarvo has joined #jruby
<lopex>
nirvdrum: I guess it doesnt matter at this point
<nirvdrum>
lopex: Did it matter at some point?
<lopex>
there's no byte thing on jvm stack
<lopex>
nirvdrum: I guess not
<lopex>
unless there's a difference on byte/int widening when masking with 0xff
<lopex>
but I thing there's not
<lopex>
*think
<chrisseaton>
bytes are inconvenient to use in Java - they never seem to cast up as I would expect - you can't call an int method with byte - you have to cast it
<lopex>
chrisseaton: but in the case of get you still get java widening
<nirvdrum>
chrisseaton: The problem here is it's really easy to forget to mask with 0xff.
<lopex>
chrisseaton: afaik the masking with 0xff has it's on case in hotspot gvn matchers
<lopex>
it's so common
<lopex>
*own
<lopex>
but yeah, not having unsigned types is a mistake in java
<lopex>
it complicates things
_djbkd has quit [Remote host closed the connection]
<lopex>
nirvdrum: and it's not end of the story
<lopex>
nirvdrum: like int -> long widening with sign
<lopex>
it will catch us too at some point with 32 codepoint encodings
<lopex>
till now, not enough data in tests :)
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 2 new commits to master: http://git.io/jXTi
<JRubyGithub>
jruby/master 0efdc52 Kevin Menard: [Truffle] Added a missing @RubiniusOnly.
<JRubyGithub>
jruby/master 66a1c8e Kevin Menard: [Truffle] Fixed String#each_byte.
JRubyGithub has left #jruby [#jruby]
erikhatcher has quit [Quit: erikhatcher]
<headius>
jiggity jig
<lopex>
headius: tell me thee DEALINE
<lopex>
*DEADLINE
kares has quit [Ping timeout: 256 seconds]
<headius>
couple weeks maybe?
<lopex>
oh
<lopex>
in that period I can catch up with onigmo
<lopex>
headius: they also added some more boyer-moore modification algorithms
<lopex>
I also want to diverge on some things
<lopex>
but need a feedback
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius pushed 3 new commits to master: http://git.io/jXLy
<JRubyGithub>
jruby/master 8c614e6 Charles Oliver Nutter: Fix literal math operations to use proper call sites.
<JRubyGithub>
jruby/master 86c68cc Charles Oliver Nutter: Avoid dispatching through IRScope to get these values.
<JRubyGithub>
jruby/master cb72352 Charles Oliver Nutter: Reduce protocol checking and reduce code for each path.
JRubyGithub has left #jruby [#jruby]
<headius>
lopex: need feedback from me?
<headius>
or planning to talk to MRI/Onigmo people?
<nirvdrum>
headius: Did you see my note earlier about String specs?
<lopex>
headius: yeah, from you, or maybe other jruby folks
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] bjfish pushed 1 new commit to master: http://git.io/jXqc
<JRubyGithub>
jruby/master 716a16c Brandon Fish: [Truffle] Moving Kernel#Array to kernel.rb.
<lopex>
headius: well, the mount of checks they have to perform for malloc failures just shadows obvious opts on one hand
<lopex>
headius: to the point, we might try to make joni throw errors on malformed strings on it's own
<lopex>
headius: and skip this preverification drama
<lopex>
onthe other hand
<lopex>
headius: that would be an invasive difference right ?
<lopex>
one complication is the preprocessing
djbkd has joined #jruby
<chrisseaton>
headius: is the asm you posted without indy right?
<headius>
without indy but using MethodHandle
<headius>
nirvdrum: no I missed it
<chrisseaton>
I didn't know MethodHandle did anything useful without indy - is it just a general function pointer even without indy?
<headius>
nirvdrum: found it... I did a pass over tags last night
<headius>
dropped about 200 across all specs
<lopex>
chrisseaton: isnt it used for method references ?
<headius>
chrisseaton: yes...in theory a faster-than-reflection function object
erikhatcher has joined #jruby
lance|afk is now known as lanceball
lanceball is now known as lance|afk
<headius>
I'm using it so I don't have to generate stub code to dispatch to jitted methods
<lopex>
headius: an opinion on the above ?
<nirvdrum>
headius: Cool.
<headius>
hmm, I probably should have turned off tiered compilation
<headius>
lopex: moving more of those checks into joni sounds great to me
<headius>
biggest complaint I've heard from other joni users is that it's brittle if you don't preverify everything
<lopex>
headius: but that wont be easy on it's own, I'm more concerned with incompatibilities
<lopex>
potential ones
lance|afk is now known as lanceball
<bjfish2>
headius i’m seeing a compilation error locally: CompiledIRMetaClassBody.java:[43,4] error: method does not override or implement a method from a supertype
<lopex>
chrisseaton: do you have any news about joni compilation ?
<bjfish2>
thanks
<chrisseaton>
lopex: nothing's happening at the moment as far as I know
<lopex>
chrisseaton: or at least any work that's accessible ?
<chrisseaton>
lopex: don't think so, sorry
<chrisseaton>
lopex: is there a particular reason you're interested now? are you trying something out or have some workload you think would benefit from compiled joni?
<lopex>
chrisseaton: there was an attempt from Oracle side once, but it was to limited for jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius pushed 1 new commit to master: http://git.io/jXWX
JRubyGithub has left #jruby [#jruby]
<JRubyGithub>
jruby/master 3b52341 Charles Oliver Nutter: Fix subclass impl of pre.
<lopex>
chrisseaton: well, I still plan to do a compiler for it, but given some attempts were done I'm interested in anything
bbrowning is now known as bbrowning_away
<lopex>
headius: is there a benchmark across different hotspots regarding different schemes like 'evolution of switch performance' ?
erikhatcher has quit [Quit: erikhatcher]
triple_b has quit [Ping timeout: 252 seconds]
x1337807x has joined #jruby
x1337807x has quit [Max SendQ exceeded]
x1337807x has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/jXzW
<JRubyGithub>
jruby/master 94e7bf9 Kevin Menard: [Truffle] Use faster version of String#size for single-byte optimizable Strings throughout.
JRubyGithub has left #jruby [#jruby]
djbkd has quit [Remote host closed the connection]
jeremyevans has quit [Ping timeout: 252 seconds]
mitchellhenke has quit [Quit: Computer has gone to sleep.]
yfeldblum has quit [Remote host closed the connection]
<headius>
I'm not sure exactly how to reproduce but I think it's your changes
<headius>
I'm seeing double requires
lanceball is now known as lance|afk
<headius>
like the RubyGems require is firing both exceptional and non-exceptional path or something
<subbu>
headius, oh hmm ..
<headius>
I don't have confirmation that it's your changes, but there's a lot of exception handling in there and I haven't touched anything I'd expect to break this
<mistergibson>
Correct me if I'm wrong, but it seems the JRUBY_OPTS has changed how you enable C ext in 9.0.0.0.pre1 - anyone know the new way?
<chrisseaton>
mistergibson: I'm afraid C extensions have gone in 9k
<mistergibson>
chrisseaton: wtf? Hrm, I really need those for my gems
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] e2 opened issue #2784: rindex() fails on strings read from binary files http://git.io/jXbt
JRubyGithub has left #jruby [#jruby]
<chrisseaton>
mistergibson: they've been gone from the 1.7.x releases for a while
<chrisseaton>
mistergibson: did they work well for you?
<subbu>
headius, is it consistently reproducible?
<mistergibson>
chrisseaton: actually the last release had it and it worked great
<mistergibson>
but the 9K doesn't ... ok, what possessed them to eliminate something that is clearly required by a lot of ruby code?
<mistergibson>
about half of my gems (at least) link to C libs
djbkd has quit [Remote host closed the connection]
yfeldblum has joined #jruby
<headius>
subbu: ok, here's what's happening...now I don't think it's your bug
<headius>
when you have an autoload for a given constant, and load the file associated with that autoload directly, when the script defines its class it triggers the autoload
<headius>
defineOrGetClassUnder is triggering autoload when it did not before
<headius>
does that sound familiar to anyone?
<subbu>
headius, not to me. "before" is when?
djbkd has joined #jruby
<headius>
good question
<headius>
I'm rather amazed all tests pass
mcclurmc has quit [Remote host closed the connection]
<headius>
but I guess reloading the same file *usually* doesn't break anything
<subbu>
you mean despite the failure?
<headius>
the failure here is that this script tries to undef a method
<headius>
so after the first execution it fails
<headius>
hmmm
iamjarvo has joined #jruby
iamjarvo has quit [Max SendQ exceeded]
momomomomo_ has joined #jruby
iamjarvo has joined #jruby
<headius>
ok, I guess I'm wrong
<headius>
it's supposed to trigger autoload...but it's supposed to detect that it's currently loading and do nothing, I guess?