mister_solo has quit [Read error: Connection reset by peer]
DomKM has quit [Quit: Connection closed for inactivity]
mitchellhenke has quit [Quit: Computer has gone to sleep.]
e_dub has joined #jruby
x1337807_ has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
zorak8 has quit [Ping timeout: 250 seconds]
tcrawley-away is now known as tcrawley
zorak8 has joined #jruby
bjfish2 has quit [Quit: bjfish2]
e_dub has quit [Quit: e_dub]
elia has quit [Quit: Computer has gone to sleep.]
e_dub has joined #jruby
e_dub has quit [Quit: e_dub]
e_dub has joined #jruby
zorak8 has quit [Read error: Connection reset by peer]
zorak8 has joined #jruby
e_dub has quit [Client Quit]
marr has quit [Ping timeout: 244 seconds]
camlow325 has quit []
maleghast has joined #jruby
havenn has joined #jruby
havenn has quit [Remote host closed the connection]
Hobogrammer has joined #jruby
e_dub has joined #jruby
e_dub has quit [Client Quit]
havenwood has quit []
baroquebobcat has joined #jruby
x1337807x has joined #jruby
x1337807x has quit [Client Quit]
bga57 has joined #jruby
phrinx_ has quit [Remote host closed the connection]
nateberkopec has quit [Quit: Leaving...]
phrinx has joined #jruby
phrinx has quit [Ping timeout: 265 seconds]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 1 new commit to test-travis-fail: http://git.io/htbJ
JRubyGithub has left #jruby [#jruby]
<JRubyGithub> jruby/test-travis-fail bd48e14 Charles Oliver Nutter: Try again.
<headius> lopex: you want to change some?
<headius> there's a few fields I am accessing from JRuby but they'd all be public
<headius> lots of logic sorta lives in between jcodings and jruby, so it's in jruby
havenwood has joined #jruby
<bga57> headius: I'm having problems with jruby on freebsd, how would I go about determining whether it's jruby or openjdk?
<headius> bga57: what's the problem?
<headius> might just be able to tell you
<bga57> For example, installing rails and then creating a new rails project, it won't create/open a sqlite database
<headius> what does it do?
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 1 new commit to test-travis-fail: http://git.io/htx0
<JRubyGithub> jruby/test-travis-fail 59fc254 Charles Oliver Nutter: ...
JRubyGithub has left #jruby [#jruby]
<headius> bga57: I'm going to have to leave soon, but if you post a bug with logs etc others will help too
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] nirvdrum pushed 6 new commits to master: http://git.io/htx9
<JRubyGithub> jruby/master 336aeb1 Kevin Menard: Moved RubyString#isCompatibleWith to StringSupport.areCompatible.
<JRubyGithub> jruby/master 9169677 Kevin Menard: [Truffle] Real implementation of RubyString#checkEncoding.
<JRubyGithub> jruby/master 3466bba Kevin Menard: Moved RubyString#addByteLists to StringSupport.addByteLists.
JRubyGithub has left #jruby [#jruby]
<bga57> I was going to make the rails app available, but I here's the first part of the trace
<bga57> arjdbc/jdbc/RubyJdbcConnection.java:453:in `init_connection' activerecord-jdbc-adapter (1.3.14) lib/arjdbc/jdbc/connection.rb:15:in `initialize' activerecord-jdbc-adapter (1.3.14) lib/arjdbc/jdbc/adapter.rb:68:in `initialize' activerecord-jdbc-adapter (1.3.14) lib/arjdbc/jdbc/connection_methods.rb:12:in `jdbc_connection' activerecord-jdbc-adapter (1.3.14) lib/arjdbc/sqlite3/connection_methods.rb:24:in `sqlite3_connection' activerecord (4.1.5) lib
<headius> that doesn't help much...I'd recommend using gist.github.com and bugs.jruby.org to file an issue
<headius> you might ask in #ruby-lang or #ruby if nobody's around here :-)
<headius> ttfn
<bga57> ok
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] nirvdrum pushed 1 new commit to master: http://git.io/hthD
<JRubyGithub> jruby/master c9a5cca Kevin Menard: [Truffle] Untagged a bunch of passing String specs.
JRubyGithub has left #jruby [#jruby]
tcrawley is now known as tcrawley-away
travis-ci has joined #jruby
<travis-ci> jruby/jruby (test-travis-fail:59fc254 by Charles Oliver Nutter): The build is still failing. (http://travis-ci.org/jruby/jruby/builds/54663390)
travis-ci has left #jruby [#jruby]
nirvdrum has quit [Ping timeout: 245 seconds]
mitchellhenke has joined #jruby
mitchellhenke has quit [Quit: Computer has gone to sleep.]
yfeldblum has quit [Remote host closed the connection]
iamjarvo has joined #jruby
mister_solo has joined #jruby
mister_solo has quit [Read error: Connection reset by peer]
calavera has joined #jruby
treehau55 has joined #jruby
<treehau55> hey all, I am trying to call init() on an SSLContext object (which I created with SSLContext.getInstance("SSL") but get TypeError: failed to coerce org.jruby.gen.InterfaceImpl938290467 to [Ljavax.net.ssl.TrustManager; Can anyone point me in the right direction here? Basically trying to use httpclient via jruby to connect over ssl, but its a self signed cert on a server I trust
digitalextremist has joined #jruby
yfeldblum has joined #jruby
bjfish2 has joined #jruby
SynrG has quit [Ping timeout: 252 seconds]
zorak8 has quit [Read error: Connection reset by peer]
mister_solo has joined #jruby
zorak8 has joined #jruby
mister_solo has quit [Read error: Connection reset by peer]
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
bjfish2 has quit [Quit: bjfish2]
baroquebobcat has quit [Quit: baroquebobcat]
yfeldblum has quit [Remote host closed the connection]
<Antiarc> Would it be at all possible to cache this?
<Antiarc> ie, how easy is it to know when a new constant is added to the hierarchy?
<Antiarc> This is responsible for the bulk of CPU time when doing JSON encoding in this app I'm profiling
<Antiarc> 3k of 6.5k seconds in the full test, in this case
<Antiarc> err, ms
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
calavera has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
calavera has joined #jruby
calavera has quit [Client Quit]
DomKM has joined #jruby
anaeem1_ has joined #jruby
rsim has joined #jruby
maleghast has quit [Ping timeout: 244 seconds]
kares has joined #jruby
e_dub has joined #jruby
rsim has quit [Quit: Leaving.]
maleghast has joined #jruby
zorak8 has quit [Ping timeout: 256 seconds]
treehau55 has quit [Quit: Page closed]
treehau55 has joined #jruby
<treehau55> I d/cd. Did anyone happen to address my question while I was away?
yfeldblum has joined #jruby
dinfuehr has joined #jruby
treehau55 has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
yfeldblu_ has joined #jruby
dinfuehr has quit [Remote host closed the connection]
yfeldblum has quit [Ping timeout: 246 seconds]
yfeldblu_ has quit [Read error: Connection reset by peer]
yfeldblum has joined #jruby
yfeldblu_ has joined #jruby
yfeldblum has quit [Ping timeout: 244 seconds]
havenwood has quit [Remote host closed the connection]
SynrG has joined #jruby
dumdedum has joined #jruby
pitr-ch has joined #jruby
dinfuehr has joined #jruby
<kares> treehau55 ;what was the q?
dinfuehr has quit [Remote host closed the connection]
drbobbeaty has joined #jruby
<mpapis> kares: <treehau55> [03:21:05] hey all, I am trying to call init() on an SSLContext object (which I created with SSLContext.getInstance("SSL") but get TypeError: failed to coerce org.jruby.gen.InterfaceImpl938290467 to [Ljavax.net.ssl.TrustManager; Can anyone point me in the right direction here? Basically trying to use httpclient via jruby to connect over ssl, but its a self signed cert on a server I trust [03:21:05]
<kares> mpapis: thanks
<kares> treehau55: that will need some more info - seems like you passed in Java proxy that seems unexpected a gist of code would be handy to see if it's not a bug
dinfuehr has joined #jruby
djellemah has joined #jruby
elia has joined #jruby
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
digitalextremist has quit [Ping timeout: 252 seconds]
digitalextremist has joined #jruby
dinfuehr has quit [Remote host closed the connection]
maleghast has quit [Ping timeout: 244 seconds]
kares has quit [Ping timeout: 264 seconds]
travis-ci has joined #jruby
<travis-ci> kares/jruby (test-ji-tuning-1_7:1c188ed by kares): The build is still failing. (http://travis-ci.org/kares/jruby/builds/54696129)
travis-ci has left #jruby [#jruby]
kares has joined #jruby
vtunka has joined #jruby
shellac has joined #jruby
mje113__ has joined #jruby
pitr-ch has quit [Quit: Leaving]
pitr-ch has joined #jruby
drbobbeaty has joined #jruby
dinfuehr has joined #jruby
dinfuehr has quit [Ping timeout: 240 seconds]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] mkristian force-pushed pack200 from 53e9ed1 to 42f65bc: http://git.io/hmH5
<JRubyGithub> jruby/pack200 ff7afd0 Christian Meier: create usual bin archive as well one where jar files are pack200 compressed...
<JRubyGithub> jruby/pack200 8987ef4 Christian Meier: added check of file size of distribtion artifacts...
<JRubyGithub> jruby/pack200 333b7ce Christian Meier: [build] generated pom.xml
JRubyGithub has left #jruby [#jruby]
benlovell has joined #jruby
pedroandrade has joined #jruby
mister_solo has joined #jruby
dinfuehr has joined #jruby
nirvdrum has joined #jruby
mister_solo has quit [Ping timeout: 256 seconds]
dinfuehr has quit [Ping timeout: 246 seconds]
benlovell has quit [Ping timeout: 250 seconds]
yfeldblu_ has quit [Ping timeout: 265 seconds]
benlovell has joined #jruby
dinfuehr has joined #jruby
marr has joined #jruby
cremes has joined #jruby
bbrowning_away is now known as bbrowning
tcrawley-away is now known as tcrawley
anaeem1_ has quit [Remote host closed the connection]
anaeem1 has joined #jruby
tcrawley is now known as tcrawley-away
bjfish2 has joined #jruby
tcrawley-away is now known as tcrawley
pedroandrade has quit [Ping timeout: 246 seconds]
kares has quit [Ping timeout: 244 seconds]
vtunka has quit [Quit: Leaving]
vtunka has joined #jruby
vtunka has quit [Remote host closed the connection]
vtunka has joined #jruby
slyphon has joined #jruby
iamjarvo has joined #jruby
enebo has joined #jruby
lance|afk is now known as lanceball
<enebo> chrisseaton: I just added bjfish to the org so he can commit. Show him the ropes if he is not already well versed
<enebo> bjfish2: I just noticed you are also on. If you commit outside of Truffle area we will probably want you to ping us or still do a pr but thanks for helping out.
<headius> enebo: build is dying in a very weird place, during a test for timeout(1) { for loop {
<headius> }
<enebo> hmmm
<headius> maven says "Killed" in the build log almost immediately
<headius> same place every time
<enebo> headius: from my commit or just a new thing?
<headius> this is what I've been investigating, only in the embedded test suites
<headius> started a few days ago when I did encoding fixes but that doesn't seem to be related
<enebo> headius: for a minute I was thnking it was my thread_poll removal
<headius> when was that?
<bjfish2> enebo okay, thanks
<enebo> headius: I guess it could be. About a month ago maybe
travis-ci has joined #jruby
<headius> seems unlikely
<travis-ci> jruby/jruby (test-travis-fail:59fc254 by Charles Oliver Nutter): The build is still failing. (http://travis-ci.org/jruby/jruby/builds/54663390)
travis-ci has left #jruby [#jruby]
<enebo> headius: we could add it back to closures and leave it off methods but I figure loop is safe
<headius> this build didn't start failing until a few days ago
viking has joined #jruby
<enebo> headius: although I do want to rewrite loop as a pure IR loop
<headius> and I'd expect it to hang rather than crash
viking has quit [Changing host]
viking has joined #jruby
<enebo> headius: although that will have a threadpoll if that opt happens
<enebo> jruby bench/core/basic_object/ivar_access_bench.rb
<enebo> ^ with -X-C it is like 2x faster than JIT
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<headius> oh really?
<headius> I must not be caching ivar accessors right
<headius> or they're invalidating every time
<headius> file a bug for me
<enebo> well I am still playing with making symbol a immut lit
<enebo> ok
mister_solo has joined #jruby
<headius> this has to be a JVM crash
<headius> I don't know how we can work around it other than trying a different JVM
<enebo> headius: I wonder if we can get access to the crash file
<enebo> headius: I am not saying it will help but I would like access to a jvm crash on travis if it generates one
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 1 new commit to test-travis-fail: http://git.io/hYKe
<JRubyGithub> jruby/test-travis-fail b43959c Charles Oliver Nutter: Try JDK8
JRubyGithub has left #jruby [#jruby]
<headius> I'm going to try 8 and see if it works, and if it does I'll go back and add dump catting to the script and run one more
<headius> I could not get it to fail when I went directly to the travis instance
<headius> I'll try that again this morning too
mister_solo has quit [Ping timeout: 246 seconds]
<enebo> headius: crash on 71?!?!
JRubyGithub has joined #jruby
<JRubyGithub> jruby/master 8e3137a Thomas E. Enebo: Make Symbol an ImmutableLiteral instead of a simple Reference....
<JRubyGithub> [jruby] enebo pushed 1 new commit to master: http://git.io/hYK1
JRubyGithub has left #jruby [#jruby]
<headius> huh?
<enebo> headius: we had IR interp crasher but I thought that only happened sometimes on 8
kares has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] enebo opened issue #2712: Micro bench is nearly 2x slower with JIT than interp http://git.io/hY6P
JRubyGithub has left #jruby [#jruby]
e_dub has quit [Quit: e_dub]
<headius> yeah
<headius> I don't remember actually
<headius> now I have access to java bug system I can file issues
<headius> nirvdrum: you had that crasher reproducible...was it on 7 or 8?
<headius> enebo: it's working with jdk8
<headius> sigh
<headius> nirvdrum: ok
<nirvdrum> But it did stop crashing at some point.
<nirvdrum> I think subbu had changed something. But I don't recall the finer points.
slyphon has joined #jruby
<nirvdrum> I can try it out later today.
lucasallan has joined #jruby
<headius> nirvdrum: did we have more info, like a bug report?
<headius> sorry if I've asked this a dozen times, but I'm finally filing a JDK bug
travis-ci has joined #jruby
<travis-ci> jruby/jruby (test-travis-fail:b43959c by Charles Oliver Nutter): The build was fixed. (http://travis-ci.org/jruby/jruby/builds/54719324)
travis-ci has left #jruby [#jruby]
dabradley has joined #jruby
<nirvdrum> headius: I wasn't ever able to distill to a reproducible case, so I didn't file the issue.
<headius> nirvdrum: ok
<headius> I wish I had more info to give
<nirvdrum> headius: Feel free to use the contents of that gist though. There is a trace in there. I just couldn't come up with the appropriate ruby to reproduce it.
<nirvdrum> I think it was a load/volume thing though.
<headius> yeah I just linked to it
<nirvdrum> It would happen in the middle of specs. But if I tried to debug, it wouldn't happen. And if I tried to run the part that blew up, in isolation, it would pass.
vtunka has quit [Quit: Leaving]
<headius> yeah, need to have a perfect environment
mitchellhenke has joined #jruby
<headius> and dammit, I still can't get this to crash without full travis run
<headius> well, time for some build trickery I guess
<headius> I wonder if I could apt-get x11 and have travis send me an xterm
mister_solo has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] k77ch7 opened issue #2713: Hash#inspect differs from MRI Ruby 2.2.0; This problem occurs if Hash has double-quoted UTF8 key. http://git.io/hYy5
JRubyGithub has left #jruby [#jruby]
<nirvdrum> It's nice you have a bug system account now. I didn't know this existed.
<headius> you have to be an openjdk author
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 1 new commit to test-travis-fail: http://git.io/hYSR
<JRubyGithub> jruby/test-travis-fail 53c41fb Charles Oliver Nutter: Return to Java 7 and try to cat crash log.
JRubyGithub has left #jruby [#jruby]
<headius> whatever that means
iamjarvo has joined #jruby
mitchellhenke has quit [Quit: Textual IRC Client: www.textualapp.com]
subbu has joined #jruby
<nirvdrum> Heh.
vtunka has joined #jruby
<headius> enebo: can I kill your build?
<enebo> headius: sure
<headius> I hope this explains similar failures on 1.7 branch
<enebo> 2015-03-17T09:04:27.828-05:00: Interpreter: I: %v_468 = copy("utf-16le")
<enebo> booo
<headius> asarih: did travis update image recently? new JDK maybe?
<headius> enebo: that's a lot of temps
<enebo> I lament that we cannot detect this as an immutable literal
<headius> what's it for?
<headius> what emits that
<enebo> headius: well our current logic cannot not use temp vars for populating a hash/array if it has something which can affect ordering
<enebo> headius: I just added a fix where I will not use a temp if it is an immutableliteral (and therefore emit out of order) since I do not see what could break
<enebo> headius: but StringLiteral is not an immutableliteral
<enebo> headius: which is my age-old gripe with Ruby :)
<headius> indeed
<asarih> headius: good morning. the last env update was early February.
<enebo> but I think we are being too conservative so I will see how this changes things
<headius> asarih: I wish I knew why this started happening then
<headius> not then...I mean I wish, then, I knew why this started happening
<enebo> asarih: That picture last night cropped ‘Hiro’ right above #yolo but in retrospect it ended up a better picture
nateberkopec has joined #jruby
mitchellhenke has joined #jruby
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<asarih> ha ha ha ha
<asarih> ↑ enebo
<enebo> asarih: It is pretty good too
<enebo> asarih: me giving rogue a 4 is odd
<asarih> enebo: indeed, it was.
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] eregon pushed 1 new commit to master: http://git.io/hY59
<JRubyGithub> jruby/master 0139e27 Benoit Daloze: [Truffle] Raise ENOTEMPTY in Dir.rmdir.
JRubyGithub has left #jruby [#jruby]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (test-travis-fail:53c41fb by Charles Oliver Nutter): The build was broken. (http://travis-ci.org/jruby/jruby/builds/54722079)
travis-ci has left #jruby [#jruby]
<eregon> headius, enebo: Hi! Do you agree on having mspec and rubyspec as git subtrees? It's very similar to current approach for usability, but it allows easy pulling from upstream and even pushing (and splitting to refine commits) if we make modifications
<headius> eregon: I think so...what's the branch where you have subtrees set up?
<headius> I want to read about subtrees but in principal I think it's good
<enebo> eregon: spec/mspec and spec/ruby are different subtrees?
<eregon> yes
<headius> yeah this looks pretty good
<eregon> they are like cp -R copies, but added bonus of tracking commit and some "git subtree --prefix spec/ruby pull --squash"
<headius> git subtree push --prefix=spec/ruby remote_name master
<eregon> headius: it's a local branch, since I kept rebasing for testing
<headius> ok
<headius> push it and I will try it out
<eregon> right, I forogt the remote in that long line, but it surely beats manual cp and such
<eregon> ok!
<headius> submodules worked great locally for me and then acted weird for everyone else
<headius> so I want to play with this a little
<eregon> yeah I started to hate them as soon as I did a git checkout other_branch
slyphon has joined #jruby
<eregon> merging would break the build as we'll new tags, is that OK?
<eregon> as we'll need new tags*
<eregon> pushed
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] eregon pushed 6 new commits to master: http://git.io/hYpc
<JRubyGithub> jruby/master 2e8fa85 Benoit Daloze: [MSpec] Remove our copy of MSpec, to use a submodule instead.
<JRubyGithub> jruby/master 218e1ba Benoit Daloze: [RubySpec] Remove our copy of spec/ruby, it will be replaced by a git submodule....
<JRubyGithub> jruby/master 815450e Benoit Daloze: Squashed 'spec/mspec/' content from commit 48dbdc3...
JRubyGithub has left #jruby [#jruby]
<headius> because tags will have to move?
<eregon> no, because I'm using ruby/rubyspec and that has a few more specs than our previous version
<headius> oh ok
<enebo> eregon: this also includes MRIs forked commits right?
<eregon> yes
<headius> nice
<eregon> and also library specs if you are interested
<headius> we're interested...since they moved out to rubysl projects they're nearly impossible to manage
<eregon> yeah, exactly :)
<eregon> though many of these specs seem to have a rather low coverage
<eregon> and some of them should be removed because it's from the 1.8/1.9 era (we only care about >= 2.0)
<asarih> headius: what is the nature of the failure on Travis?
<asarih> headius: one thing that's different is that JRuby tests run on Docker (and EC2).
<headius> asarih: that didn't change in the past few days though
<headius> this just started failing end of last week
<headius> no obvious change on our end
<headius> enebo: no crash log it appears
<enebo> headius: hmmm weird
<asarih> :-(
<headius> may be that running several levels of process under maven confounded the crash reporter
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 1 new commit to test-travis-fail: http://git.io/hOvN
<JRubyGithub> jruby/test-travis-fail dedc6bc Charles Oliver Nutter: One more try.
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] bjfish opened pull request #2714: [Truffle] Updates to Array#{map, map!}. (master...truffle_array_map_collect) http://git.io/hOfJ
JRubyGithub has left #jruby [#jruby]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:0139e27 by Benoit Daloze): The build has errored. (http://travis-ci.org/jruby/jruby/builds/54723489)
travis-ci has left #jruby [#jruby]
<headius> travis blip
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] mkristian created strict-base64-encoding (+1 new commit): http://git.io/hOTv
JRubyGithub has left #jruby [#jruby]
<JRubyGithub> jruby/strict-base64-encoding 9a88ecf Christian Meier: implements strict base64 encoding...
mister_solo has quit [Ping timeout: 245 seconds]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] mkristian opened pull request #2715: implements strict base64 encoding (jruby-1_7...strict-base64-encoding) http://git.io/hOLi
JRubyGithub has left #jruby [#jruby]
shellac has quit [Remote host closed the connection]
shellac has joined #jruby
calavera has joined #jruby
havenwood has joined #jruby
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:0139e27 by Benoit Daloze): The build passed. (http://travis-ci.org/jruby/jruby/builds/54723489)
travis-ci has left #jruby [#jruby]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:247bba3 by Benoit Daloze): The build has errored. (http://travis-ci.org/jruby/jruby/builds/54725241)
travis-ci has left #jruby [#jruby]
<headius> eregon: that looks like untagged failures...did you tag for non-truffle?
<eregon> no, not yet, work in progress
<headius> ahh ok
<headius> asarih: that log limit is a new thing...but it prevents us from seeing full rubyspec output
<eregon> I welcome any help :) right now working on the truffle part as I know what we should fail or not
<headius> we'll turn off specdoc but that's a little onconvenient
<headius> in
<headius> eregon: ok
<asarih> headius: log limit is not new. it's always been there.
<headius> hmm
<headius> asarih: it kills the build, yes?
<headius> maybe added specs are pushing us over now
<asarih> I *think* so.
<headius> eregon: killing build since it will fail anyway and I'm trying to settlt this JDK7 issue
<headius> settle
<eregon> ok!
<eregon> headius: your failure wasnt happening on the ruby:spec:fast build I hope?
<headius> eregon: my crash?
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 1 new commit to master: http://git.io/hO4d
<JRubyGithub> jruby/master 6d8b0eb Charles Oliver Nutter: Disable specdoc output since we exceed travis limit now.
JRubyGithub has left #jruby [#jruby]
<headius> crash was happening in the complete-jar runs of test:jruby
mcclurmc has quit [Remote host closed the connection]
mcclurmc has joined #jruby
slyphon has joined #jruby
e_dub has joined #jruby
camlow325 has joined #jruby
erikhatcher has joined #jruby
<asarih> headius: for the failing builds on Docker, can you try removing `sudo: false` and see if it persists? (of course, you should reduce the matrix size)
travis-ci has joined #jruby
<travis-ci> jruby/jruby (test-travis-fail:dedc6bc by Charles Oliver Nutter): The build was fixed. (http://travis-ci.org/jruby/jruby/builds/54726569)
travis-ci has left #jruby [#jruby]
baroquebobcat has joined #jruby
<headius> that wasn't supposed to pass
<headius> oh nevermind, passed because build was not last command
dinfuehr_ has joined #jruby
<headius> asarih: sure
rcvalle has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 1 new commit to test-travis-fail: http://git.io/hOov
<JRubyGithub> jruby/test-travis-fail 6bdfd35 Charles Oliver Nutter: Try removing sudo: false.
JRubyGithub has left #jruby [#jruby]
dinfuehr has quit [Ping timeout: 255 seconds]
mister_solo has joined #jruby
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:247bba3 by Benoit Daloze): The build was canceled. (http://travis-ci.org/jruby/jruby/builds/54725241)
travis-ci has left #jruby [#jruby]
dinfuehr_ has quit [Remote host closed the connection]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] lucasallan pushed 1 new commit to master: http://git.io/hOMV
<JRubyGithub> jruby/master 7c3b89b Lucas Allan Amorim: [Truffle] - More Random specs passing.
JRubyGithub has left #jruby [#jruby]
e_dub has quit [Quit: e_dub]
e_dub has joined #jruby
e_dub has quit [Client Quit]
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
e_dub has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] chrisseaton pushed 2 new commits to master: http://git.io/hOxh
<JRubyGithub> jruby/master 9092267 Brandon Fish: [Truffle] Fixing Array#[]= for Fannduch benchmark.
<JRubyGithub> jruby/master 3dafe3f Chris Seaton: Merge pull request #2709 from bjfish/truffle_array_fannduch_fix...
JRubyGithub has left #jruby [#jruby]
benlovell has quit [Ping timeout: 245 seconds]
<headius> enebo: full spec run takes a *looooong* time
<headius> Regexp with character classes doesn't match Unicode Mongolian vowel seperator characters with [[:blank:]
<headius> thank goodness that's getting tested
<enebo> headius: hah
<headius> there's a couple kwrest failures in updated specs
<enebo> headius: full spec with jit? :)
<enebo> headius: no doubt I have the fix for those if they involve destructive behavior
<headius> I'm just going to tag off what's failing
<enebo> headius: cool
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] eregon pushed 9 new commits to master: http://git.io/h3UL
<JRubyGithub> jruby/master b10672a Benoit Daloze: [Truffle] Basic File.realpath.
<JRubyGithub> jruby/master 9a17acf Benoit Daloze: [Truffle] Also print the value of a boolean in UnsupportedSpecializationException.
<JRubyGithub> jruby/master 65e840a Benoit Daloze: [Truffle] Fix Time failures.
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] eregon pushed 2 new commits to master: http://git.io/h3U3
<JRubyGithub> jruby/master 0b1d0bc Benoit Daloze: Squashed 'spec/ruby/' changes from 0c3c64b..63695c3...
<JRubyGithub> jruby/master 07d7f57 Benoit Daloze: Merge commit '0b1d0bc6dfaa085e78e8c47c739c5200ab1e1218'
JRubyGithub has left #jruby [#jruby]
<headius> enebo: looks like they're related to required args
<eregon> headius: tried a subtree split and a subtree pull, it works nicely, with pull being quite a bit easier and faster
<eregon> have a look at git history, it looks quite nice
<headius> ok thanks
<headius> I'm reviewing failures to tag off
<headius> "The break statement in a captured block when the invocation of the scope creating the block is still active raises a LocalJumpError when invoking the block from the scope creating the block"
e_dub has quit [Quit: e_dub]
<eregon> ahah, ParseError on that
<headius> no doubt
<eregon> how many failures do you have on jruby?
<headius> it was at 140-something when it started to get into library specs
<headius> most of them look like edge cases
<eregon> yeah, that's typically what the MRI commits on rubyspec contains, the specs for new edge cases
pitr-ch has quit [Quit: Leaving]
<headius> they'll be good to fix at some point, but I'm mostly just tagging
<eregon> my technique right now is to run with -V, so I can run per failing file
e_dub has joined #jruby
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:7c3b89b by Lucas Allan Amorim): The build failed. (http://travis-ci.org/jruby/jruby/builds/54738848)
travis-ci has left #jruby [#jruby]
rsim has joined #jruby
benlovell has joined #jruby
lucasallan has quit [Remote host closed the connection]
<headius> asarih: removing sudo:false worked!
<headius> what made you think to try that?
<asarih> headius: seems that we are having problem with JVM on Docker.
<asarih> seems to be a recent development
<asarih> not sure why.
<asarih> or what would explain this difference, since we haven't touched it in a month or so.
<headius> well I'm glad there's some sort of explanation
<headius> asarih: is it possible to run only some builds in Docker?
<headius> I mean, only some parts of the matrix
<asarih> headius: unfortunately not.
<headius> ok
<headius> I can make these suites run under 8 for now...it's not a big deal
<asarih> ok.
vtunka has quit [Quit: Leaving]
subbu has quit [Quit: Ex-Chat]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 3 new commits to master: http://git.io/h3s7
<JRubyGithub> jruby/master 2c8d0e8 Charles Oliver Nutter: Tag off new failures in RubySpec.
<JRubyGithub> jruby/master 771e628 Charles Oliver Nutter: Fix verbose warning for &arg.
<JRubyGithub> jruby/master 6a83754 Charles Oliver Nutter: Switch crashing builds to JDK8 and remove from failures.
JRubyGithub has left #jruby [#jruby]
<headius> asarih: I won't touch the master builds that have this issue, in case you need them to reproduce something
maleghast has joined #jruby
Hobogrammer has quit [Ping timeout: 264 seconds]
<headius> this is the most recent that didn't fail for any other reasons: https://travis-ci.org/jruby/jruby/builds/54723489
<headius> it appears that the subprocess JVM just goes away mid-run
<headius> maven reports "Killed" and that's it
camlow32_ has joined #jruby
<asarih> ok. so that's typically OOMKiller.
<headius> I believe you, but it crashed at the exact same test every time
<headius> or at least, it did when I was investigating
<asarih> WAT
<asarih> that's curious..
<headius> only on oraclejdk7/openjdk7, only with sudo:false, only headless travis (not on debug instance)
<headius> it's a great mystery
camlow325 has quit [Ping timeout: 256 seconds]
lucasallan has joined #jruby
pjammer has joined #jruby
phrinx has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] lucasallan pushed 2 new commits to master: http://git.io/h3lp
<JRubyGithub> jruby/master 1254dfa Lucas Allan Amorim: [Truffle] - Adds Kernel#srand
<JRubyGithub> jruby/master 661b414 Lucas Allan Amorim: [Truffle] - Move code from rubinius/common/thread.rb to rubinius/api/kernel/common/thread.rb
JRubyGithub has left #jruby [#jruby]
<headius> asarih: seems like a similar problem with openjdk6 on 1.7 branch
<headius> again, just started in the past week
<asarih> o_O
<headius> enebo: AST tool busted
<headius> ast --ir raises an error, type coercion issue...probably some API changed
<headius> TypeError: failed to coerce org.jruby.ir.interpreter.BeginEndInterpreterContext to org.jruby.ir.IRScope
<headius> I'll see what I can do with it
camlow32_ has quit [Remote host closed the connection]
shellac has quit [Quit: Ex-Chat]
maleghast has quit [Quit: Textual IRC Client: www.textualapp.com]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (jruby-1_7:af940b1 by Charles Oliver Nutter): The build is still failing. (http://travis-ci.org/jruby/jruby/builds/54611871)
travis-ci has left #jruby [#jruby]
mister_s_ has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 1 new commit to jruby-1_7: http://git.io/h3zZ
<JRubyGithub> jruby/jruby-1_7 297c953 Charles Oliver Nutter: Use JDK8 for these three...they are affected by JVM+Travis issues.
JRubyGithub has left #jruby [#jruby]
elia has quit [Quit: Computer has gone to sleep.]
mister_solo has quit [Ping timeout: 245 seconds]
<headius> asarih: is there a bug report I can add to, or should I file one?
camlow325 has joined #jruby
<asarih> none that I'm aware. having one is a good idea.
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] chrisseaton pushed 2 new commits to master: http://git.io/h3gx
<JRubyGithub> jruby/master 98d38ce Brandon Fish: [Truffle] Moving Array#to_s to array.rb
<JRubyGithub> jruby/master 02a4da2 Chris Seaton: Merge pull request #2711 from bjfish/truffle_array_to_s...
JRubyGithub has left #jruby [#jruby]
<headius> asarih: ok
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] eregon pushed 2 new commits to master: http://git.io/h3VH
<JRubyGithub> jruby/master 28e8a6d Benoit Daloze: [Truffle] Tag failing core specs.
<JRubyGithub> jruby/master edafd13 Benoit Daloze: [Truffle] Add Thread#join(nil).
JRubyGithub has left #jruby [#jruby]
<eregon> headius: how is tagging going? can I help?
<headius> I pushed mine already
<eregon> sweat, thanks! I fixed truffle's build so now we should get green-ish again
<eregon> sweet*
<headius> eregon: ok, thanks for getting us updated and on subtree
rsim has quit [Read error: No route to host]
rsim has joined #jruby
dumdedum has quit [Quit: foo]
lucasallan has quit [Remote host closed the connection]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] chrisseaton pushed 1 new commit to truffle-kwargs: http://git.io/h3M3
<JRubyGithub> jruby/truffle-kwargs 44060f3 Chris Seaton: Merge branch 'master' into truffle-kwargs
JRubyGithub has left #jruby [#jruby]
<headius> hmm, a couple failures only on travis
<headius> I guess I'll let the build settle and then have a look
<headius> ahh ruby_bug
<headius> we need to get MRI to use tags instead of ruby_bug
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:6a83754 by Charles Oliver Nutter): The build was canceled. (http://travis-ci.org/jruby/jruby/builds/54747599)
travis-ci has left #jruby [#jruby]
zorak8 has joined #jruby
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:6a83754 by Charles Oliver Nutter): The build was canceled. (http://travis-ci.org/jruby/jruby/builds/54747599)
travis-ci has left #jruby [#jruby]
<chrisseaton> what is ruby_bug?
<headius> used to mark rubyspecs that are accepted as bugs by ruby-core but not yet in a release
<headius> they don't run...but only on MRI
<chrisseaton> oh so it goes right into the specs source code?
<headius> yeah
<headius> look at constants_spec for example
<chrisseaton> we should be able to automatically generate tests from that I guess
<headius> there's some new constant behavior for 2.3 that's marked ruby_bug right now
<chrisseaton> generate tags I mean
<headius> yeah
zorak8 has quit [Ping timeout: 265 seconds]
<headius> they should run and tag as normal for us
nullsign has left #jruby ["Textual IRC Client: www.textualapp.com"]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:6a83754 by Charles Oliver Nutter): The build was canceled. (http://travis-ci.org/jruby/jruby/builds/54747599)
travis-ci has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 3 new commits to master: http://git.io/h3NM
<JRubyGithub> jruby/master 123c379 Charles Oliver Nutter: Fix ast IR printing for recent changes.
<JRubyGithub> jruby/master 288b868 Charles Oliver Nutter: JIT runs should not use --dev.
<JRubyGithub> jruby/master 3ea1da0 Charles Oliver Nutter: Tag additional failures and remove bogus tag files.
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] chrisseaton pushed 1 new commit to master: http://git.io/h3x9
<JRubyGithub> jruby/master 8e37042 Chris Seaton: [Truffle] Fix location of Rubinius and rubinius-core-api thread.rb code.
JRubyGithub has left #jruby [#jruby]
benlovell has quit [Ping timeout: 244 seconds]
<headius> catching up travis builds everyone...hopefully everything gets back to green now
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:661b414 by Lucas Allan Amorim): The build was canceled. (http://travis-ci.org/jruby/jruby/builds/54749643)
travis-ci has left #jruby [#jruby]
<eregon> headius, chrisseaton: but ruby_bug is quite a bit more than just a tag, so it's not obvious to migrate
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] nirvdrum opened issue #2716: rb_cmpint difference from MRI http://git.io/h3pX
JRubyGithub has left #jruby [#jruby]
<headius> eregon: yeah, it's also annoying
<headius> I wish it had been removed back when we said it should be removed
<eregon> headius: BTW, you added a commend on a truffle tag for def_spec, not sure if intended but it certainly does not hurt :)
<nirvdrum> I'll fix #2716 unless anyone has a valid reason for our different behavior.
<headius> oops
<eregon> headius: that's the moment MRI forked rubyspec
<eregon> essentially, if they are just tags: there is no link to find the original bug/ongoing discussion, there is no version check in tags (MRI needs to keep ~4 release at the same time) and running with old released MRI will have failures
<headius> eregon: you can put comments in tags, though, and tags should live in MRI's repo, not rubyspec
<eregon> but then you need to keep tags per branch at least
robbyoconnor has quit [Ping timeout: 264 seconds]
<headius> yes, I don't see a problem with that
<eregon> which for them is probably some hassle in svn
<headius> and you can't run old MRI with new specs anyway
<headius> could be...I wish they'd hurry up and move to git
<eregon> not too old, but norammly anything > 2.0.0p0 should run the specs and normally pass
<eregon> (right now there a couple known failures on 2.0.0)
rsim1 has joined #jruby
<headius> eregon: these constant specs should just be in a version guard, not a ruby_Bug
<eregon> headius: issue nb tends to be useful when you have a new failures for other implementations, don't you think?
<eregon> which ones?
<headius> ones I just tagged
<eregon> yeah I wrote those
<eregon> not sure what was appropriate
<headius> they're ruby_bug and mention a bug number and ruby version
<headius> that should just be a version guard I would say
<headius> are they actually fixed on MRI head?
travis-ci has joined #jruby
<travis-ci> jruby/jruby (jruby-1_7:297c953 by Charles Oliver Nutter): The build has errored. (http://travis-ci.org/jruby/jruby/builds/54751720)
travis-ci has left #jruby [#jruby]
<eregon> but then other impl could consider behavior of MRI before that to be spec
<eregon> but behavior was always intended to be the now-specified one
<eregon> yes, they are
<headius> then it should just be bare :-)
<headius> and MRI should tag it
<headius> does MRI have any tags?
<eregon> no
<headius> eregon: I guess my point is that if there's a bug in MRI, they can use ruby_bug...if there's a bug in JRuby, we can't
<headius> so it doesn't make sense to me
<eregon> yeah, and you don't mind having to manually search on bugs.ruby-lang.org to find why the behavior changed?
<headius> well, anything you'd put in a ruby_bug line could go in the specdoc or in a comment
<eregon> yeah, but for them if there is sth in RubySpec which does not pass on MRI: it's a buggy spec or it's a ruby bug and then be annotated as such
rsim has quit [Ping timeout: 265 seconds]
<headius> it's not a ruby bug, it's an MRI bug
<headius> and so it should be a tag :-)
<eregon> yeah an MRI bug
<headius> it wouldn't make sense for us to have a bunch of jruby_bug guards instead of tags, right?
<eregon> yeah, it's a good alternative, just lose quite some info which might be valuable
baroquebobcat has quit [Quit: baroquebobcat]
<eregon> no, this is special treatment for the reference implementation
<headius> that's hard to argue against
<eregon> people expect behavior of MRI is spec, and so I do value a bit having these guards to say "this is known bug"
<headius> I'm not religious about this either, I just fine ruby_bug annoying because we have to do extra work for them
baroquebobcat has joined #jruby
<eregon> like what?
<headius> work MRI doesn't have to do, even if they're MRI's bugs
temporalfox has quit [Quit: Textual IRC Client: www.textualapp.com]
<headius> well, we have to tag and they don't
<eregon> ah, ok
<headius> it also hurts our (and truffle+jruby) pass/fail rates compared to MRI
<headius> MRI doesn't even pass all of MRI's specs, but you can't tell because of ruby_bug
<headius> etc
<eregon> yeah, pass rate of MRI should definitely counts its ruby_bug guards
<eregon> also, ruby_bug is used in cases where tags don't suffice, like SEGV etc
<eregon> but these are rather rare since it would need to be parse-time SEGV
<headius> yes
SynrG has quit [Ping timeout: 265 seconds]
<headius> and it wouldn't fix that anyway :-)
<headius> because it still has to parse
<eregon> yeah, bad point
<headius> heheh
<headius> anyway...I've said my piece and tagged off the bugs :-)
<eregon> I am not sure what's best, I think MRI will be reluctant to have tags on all their branches and maintain them
<eregon> and I do worry a bit about the potentially less-easy to access information
<headius> damn, jdk8 didn't fix 1.7 branch build problems
<eregon> but on the other hand, ruby_bug is indeed special case and there should be a less intrusive mechanism like tags
<headius> I'm going to have to switch it to vm-based travis
<headius> eregon: to be fair there's probably still "deviates_on :jruby" in the specs
rsim1 has quit [Read error: Connection reset by peer]
<eregon> yes, I think that should totally disappear
<headius> though usually those are spec grey areas
<eregon> it should simply not be specified if this is impl-defined
<headius> like floating point behavior at the fringes
<headius> perhaps true
<eregon> or be specified with alternatives
<headius> yeah
<eregon> yeah I think [...-308, ...-324].should include Float::MIN
yfeldblum has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 2 new commits to jruby-1_7: http://git.io/hsIC
<JRubyGithub> jruby/jruby-1_7 df6bae2 Charles Oliver Nutter: Revert "Use JDK8 for these three...they are affected by JVM+Travis issues."...
<JRubyGithub> jruby/jruby-1_7 14d1d29 Charles Oliver Nutter: Try turning off sudo:false to fix build issues.
JRubyGithub has left #jruby [#jruby]
mister_s_ has quit [Ping timeout: 240 seconds]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 1 new commit to master: http://git.io/hsLz
<JRubyGithub> jruby/master 8307787 Charles Oliver Nutter: Rogue c.
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius pushed 1 new commit to master: http://git.io/hstg
<JRubyGithub> jruby/master 1da022c Charles Oliver Nutter: Oops, meant this to go into JRuby tags.
JRubyGithub has left #jruby [#jruby]
lanceball is now known as lance|afk
<headius> ok, lunch break while these builds catch up
<headius> asarih: issue filed
lucasallan has joined #jruby
lucasallan has quit [Ping timeout: 250 seconds]
SynrG has joined #jruby
calavera has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] chrisseaton closed pull request #2714: [Truffle] Updates to Array#{map, map!}. (master...truffle_array_map_collect) http://git.io/hOfJ
JRubyGithub has left #jruby [#jruby]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (jruby-1_7:14d1d29 by Charles Oliver Nutter): The build has errored. (http://travis-ci.org/jruby/jruby/builds/54761813)
travis-ci has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub> jruby/master 714712e Kevin Menard: Moved String#<=> helper methods to StringSupport.
<JRubyGithub> [jruby] nirvdrum pushed 2 new commits to master: http://git.io/hsXQ
<JRubyGithub> jruby/master a416467 Kevin Menard: [Truffle] Completed String#<=>.
JRubyGithub has left #jruby [#jruby]
<nirvdrum> String#<=> was a lot more involved than I expected.
<chrisseaton> nirvdrum: can you explain CmpIntNode? Something to do with the returned value from <=>?
<enebo> nirvdrum: I also question your use of grammar
<enebo> :)
<nirvdrum> chrisseaton: Yeah. Basically an object can return anything from <=>. CmpIntNode tries to limit it to {-1, 0, 1} if it can.
<nirvdrum> enebo: Which part?
<enebo> Although Strings are comparable so I sort of get why you changed it from is
<enebo> nirvdrum: idiomatically I don’t see predicates which start with the word ‘are'
<enebo> nirvdrum: but I don’t mind the change…I just thought I would poke :)
robbyoconnor has joined #jruby
robbyoconnor has quit [Client Quit]
<nirvdrum> Oh, I gotcha. I can change that. isWhatever looks awkward to me on static methods.
<enebo> nirvdrum: yeah I don’t care tbh
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] enebo pushed 1 new commit to master: http://git.io/hsS8
<JRubyGithub> jruby/master a6262fb Thomas E. Enebo: ImmutableLiterals can be emit'd as Operands out of order in simple cases
JRubyGithub has left #jruby [#jruby]
<enebo> nirvdrum: I just reflected that I have not seen this pattern before
<nirvdrum> If one of the strings were the receiver, I would've kept it.
<enebo> nirvdrum: It makes sense to me it is just new to me
<nirvdrum> enebo: I might be doing it wrong. It's not a bean, so I really don't know what the rule of thumb is.
<enebo> nirvdrum: I doubt there is a rule since as you just said this is not trying to be a JavaBean or something like that
<enebo> nirvdrum: especially since it is static
<nirvdrum> enebo: Do I need to set a milestone on this? https://github.com/jruby/jruby/issues/2716
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] nirvdrum pushed 1 new commit to master: http://git.io/hs9P
<JRubyGithub> jruby/master 893aefe Kevin Menard: cmpint should return {-1, 0, 1} for Fixnums....
JRubyGithub has left #jruby [#jruby]
<enebo> nirvdrum: yes
<enebo> nirvdrum: or my script slurps it up
<nirvdrum> 9.0.0.0.pre2?
<enebo> nirvdrum: yeah
<enebo> nirvdrum: I like the scripts limitation since it forces me to review all untargeted resolved issues
lucasallan has joined #jruby
<chrisseaton> nirvdrum: got it
<nirvdrum> chrisseaton: If this was the wrong approach, let me know. I was going to do it as a static utility method at first. But doing it as a node seemed natural and it helped me avoid a bunch of instanceof calls.
<chrisseaton> nirvdrum: yeah it's good
lucasallan has quit [Ping timeout: 264 seconds]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:1da022c by Charles Oliver Nutter): The build has errored. (http://travis-ci.org/jruby/jruby/builds/54762525)
travis-ci has left #jruby [#jruby]
erikhatcher has quit [Quit: erikhatcher]
<headius> enebo: fixing non-indy variable accessor caching didn't fix that bench...there's something egregious in the JVM6 compiler that's doing massive allocation
<enebo> headius: fwiw the symbol retrieval itself is much slower than interp too
<enebo> headius: since you need to ask encoding table and symbol table per ivar access
<headius> hmmm yeah
<enebo> headius: versus single site cache
<headius> chrisseaton: are your CI benchmarks running jruby with AND without indy?
<enebo> headius: I doubt that is all of that though
<headius> it would be helpful if we had both runs so I can tell what I'm doing badly in JVM6 JIT
<chrisseaton> headius: we turned off the non-indy version, as it was consuming so much time
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] chrisseaton pushed 1 new commit to master: http://git.io/hsFh
<JRubyGithub> jruby/master acd4108 Chris Seaton: [Truffle] Switch to using library rather than RubySL specs.
JRubyGithub has left #jruby [#jruby]
<headius> chrisseaton: ahh, hmm
<chrisseaton> headius: I can probably turn it back on if you want
<chrisseaton> headius: maybe run it once a week
<headius> chrisseaton: if you could at least get one run of it
camlow32_ has joined #jruby
<headius> it shouldn't be more than a couple times slower than indy version
<chrisseaton> yeah I'll turn it back on then - you can also ask me to do a run every now and again
<headius> but I need to know where it is
<headius> sure, either way
<headius> thanks
calavera has joined #jruby
<headius> heh...the top 20 allocation sites are all byte[]
<headius> that's probably the symbol caching
camlow325 has quit [Ping timeout: 244 seconds]
subbu has joined #jruby
tcrawley is now known as tcrawley-away
<headius> enebo: hah, I didn't look at the bench...it doesn't even do literal ivar access
<headius> they're all instance_variable_get
<headius> so it's the symbols
<enebo> hah yeah sorry reflective api
<enebo> yeah definitely symbol issues
<enebo> but I did not expect the allocation
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] bjfish opened issue #2717: [Truffle] File.realpath error http://git.io/hsAv
JRubyGithub has left #jruby [#jruby]
<enebo> I expected lots of table lookups
<enebo> unless it is the encoding lookup
<chrisseaton> headius: I'll do one of those big benchmarking reports for you - that includes indy and non-indy, and compares against 17
<enebo> perhaps byte[] from getName causes some other byte[] to be makde
jwinter_ has joined #jruby
<enebo> headius: also I think getName from Encoding is never ever an alias so we might be able to change the helper impl
<nirvdrum> Why does Encoding#getName return a byte[]? I've been really lazy and just used toString at times.
<enebo> headius: we should use an int in jcodings per encoding and screw all this byte[] lookup
<headius> nirvdrum: only real reason is because it was a port from C
<headius> I don't know anywhere in jcodings or joni that need it to be byte[]
<enebo> haha funny
<enebo> nirvdrum: I used byte[] directly with the encodingdb for IR persistence
<headius> enebo: yeah, a lot of the stuff around encoding naming, id, index is relics from MRI data structures we don't have or need
<enebo> nirvdrum: but it bugs me I have to store a byte[] into persisted stream
<enebo> nirvdrum: I have no constant pool in persisted format so this sort of sucks
<headius> oh jeez
<headius> enebo: IRRuntimeHelpers.getJCodingsEncoding is used by JVM6 JIT and calls through String.getBytes
<enebo> hahaha
<enebo> yes!
<headius> I assume it's to provide encoding for the bytelist to become a symbol
<enebo> headius: IRReaderStream has better impl decodeEncoding
<nirvdrum> enebo: I'm able to avoid it in most cases because we have a RubyEncoding wrapper and a table lookup. But sometimes I just want to debug.
<enebo> nirvdrum: yeah ni that case new String(bytes) is no big deal
<nirvdrum> Yeah. I'm lazy and just want to expand the object in the debugger.
<enebo> headius: EncodingDB.getEncodings().get(encodingName).getEncoding();
<enebo> encodingName is byte[]
<enebo> headius: since you are storing byte[] from encoding itself you can ignore alias case
<headius> right, that's the problem though
<headius> I am storing encoding name as String in constant pool
<headius> I just need to fix this to cache all these things properly
<headius> encodings, symbols, bytelists
<enebo> headius: oh yeah you need a byte[] pool
<enebo> headius: although I was also thinking a higher pool would be nice
<headius> that's basically RuntimeCache but I'm not using it yet in 9k
<headius> I might never
<enebo> headius: aload’ing a symbol when the symbol is in a loop would be nice :)
<headius> caching is all just synthetic methods and fields generated into the compiled code
<headius> we'll see how bad that is for meatspace
<headius> enebo: IR should be able to do that already
<headius> if there's N of the same symbol in a method there could be only one retrieval
<enebo> headius: except we sort of have the opposite issue
<enebo> headius: we will propagate literals to where they are used and not indirect through a temp
<enebo> headius: so the same symbol might be referenced as independent operand n times
<enebo> headius: in a sense this is almost the exact opposite of what you want but I think this is one layer where perhaps the JIT should decide how to cache constant values
<enebo> subbu: ^
anaeem1 has quit [Remote host closed the connection]
<headius> enebo: mmm right
* subbu is hoping for a summary :)
<headius> well in the JIT, any given symbol is only going to be cached once per compilation unit, regardless of how many places it is accessed
<enebo> subbu: Hopefully that is enough context…basically we const prop to sites in IR which is good but it means JIT needs to make some cache retreival for those constants
<headius> so I'm already doing that basically
anaeem1_ has joined #jruby
<enebo> subbu: headius’s explanation is probably the more concise of the two explanations…TAPESTRY
<enebo> to me it makes sense this is something bytecode generator would do separate from IR but I guess I am hoping for true wisdom :)
<subbu> will be here in a little bit .. discussing some edge case wikitext crap in the parsoid channel ..
<headius> hah, burned by operator precedence in ternary
<headius> that hasn't happened in a while
<enebo> headius: ternaries all the way down?
<headius> String methodName = write ? "ivarSet" : "ivarGet" + getClassData().callSiteCount.getAndIncrement() + ":" + JavaNameMangler.mangleMethodName(name);
<headius> WRONGO
<enebo> hahah
anaeem1_ has quit [Remote host closed the connection]
<nirvdrum> And now you know why for years I refused to use it. Now I reluctantly do.
<headius> ok, I have added caching for ivar accessors, encodings, and symbols
<headius> the latter two were just an oversight
anaeem1_ has joined #jruby
<nirvdrum> My business partner at my first startup spent a solid week trying to track down a firmware issue. It turned out to be a multiline macro using nested ternary ops that wasn't paired properly. Good times.
<enebo> headius: cool. I wonder how much of the 1.7 gap will be made up here
<headius> have you mostly been testing JVM6 JIT perf?
<enebo> nirvdrum: macro + ternary
<enebo> headius: I virtually never time JIT at all
camlow32_ has quit [Remote host closed the connection]
<enebo> headius: I am too concerned with startup but I sometime run JIT with JVM6
<enebo> headius: realistic mixed mode to me
<headius> yeah
<headius> I usually am running without any flags but occasionally --dev
<enebo> yeah I spend most time with —dev
<headius> ok, perf is back to where it should be, just slightly off from indy
<headius> glad you reported this
<enebo> and now I am also doing —dev -Xir.jit.threshold=-1
<enebo> So I am not even hitting full IR interps that often (other than hitting test suites)
<headius> appears to be maybe 10% faster than 1.7 too
<enebo> This is actually a problem to be addressed since our test suites are running -X-C mixed interpreters
<enebo> headius: great
donValentin has quit [Quit: donValentin]
anaeem1_ has quit [Ping timeout: 240 seconds]
<chrisseaton> headius: do you want me to include non-JIT performance in this report?
<enebo> startupinterp -> fullinterp; startupinterp -> non-indy JIT; startupinterp -> indy JIT
<enebo> and I guess startupinterp only
<enebo> heh
<enebo> too many permutations
lucasallan has joined #jruby
<enebo> this is ignoring all the indy flags themselves
baroquebobcat has quit [Quit: baroquebobcat]
<headius> chrisseaton: not my area :-)
e_dub has quit [Quit: e_dub]
<headius> it would be nice but it's nowhere near as important as tracking JIT perf
camlow325 has joined #jruby
<enebo> chrisseaton: for interp perf -X-C is fine but it is largely only important for startup performance
camlow325 has quit [Remote host closed the connection]
<subbu> headius, enebo back now.
anaeem1 has joined #jruby
e_dub has joined #jruby
<headius> chrisseaton: if we could get one run alongside the others it would be nice to have for reference in short term
<enebo> subbu: hopefully you have the scrollback to see the discussion and our two summaries
<headius> we'll see how long it takes to do interp and JVM6 JIT and then decide how often to run them, ok?
<enebo> headius: Is it madness if I try and push Signature through and kill Arity?
<headius> thank you for hooking them up
<headius> enebo: not at all
<enebo> headius: kwargs processing would really like to be able to get one type of class for this
<headius> it's a public API though and we probably can't remove it entirel
lucasallan has quit [Remote host closed the connection]
<headius> at least not without auditing some exts
<enebo> headius: so long as we make a toArity() on Signature I think that is good right?
<subbu> chrisseaton, IMO, it would be good to include it.
<headius> enebo: oh sure, I thought you meant deleting it from the codebase
<headius> killing *our* uses of it should be 100% ok
<enebo> headius: no I just want one class to look at from within IR
<subbu> headius, enebo so .. caching is not exposed in the IR .. so, there is nothing that can be done about optimizing cached access.
<headius> subbu: sure, that's ok for now too
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius closed issue #2712: Micro bench is nearly 2x slower with JIT than interp http://git.io/hY6P
JRubyGithub has left #jruby [#jruby]
<headius> JIT can be smarter about it
<enebo> subbu: so the guiddance I was wondering about is: Should IR ever do that?
camlow325 has joined #jruby
<headius> asarih: I did end up having to turn off sudo:false on jruby-1_7
<headius> went green the next build
<headius> :-(
<enebo> subbu: I mean in a sense making a single unique temp per constant and setting before first use would be a single alloc cache
<enebo> subbu: that seems wrong to me for constant values though
<enebo> subbu: or it does not seem like something which should be explicit
<enebo> subbu: So I am just wondering what your thoughts are
<subbu> since literal access is in reality a more complex operation invoving other runtime structures .. it seems okay to expose it later on (much like method lookup and caching of it, for ex.).
<enebo> subbu: for non-side-effecting values I could see doing something like this (although in Ruby what is that :) )
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] headius opened issue #2718: JVM6 JIT caches duplicate values multiple times per compile unit http://git.io/hGmY
JRubyGithub has left #jruby [#jruby]
<enebo> subbu: yeah and I can see that caching impl class sort of thing as part of IR
<enebo> subbu: but that does not feel the same as a Fixnum in my head
<headius> hey specs are all back to green in latest build
<headius> maybe we'll be green across the board when it's done
<subbu> because you have a global cache already for the small-valued fixnums.
<headius> enebo: time to stat talking pre2 schedule I think
<enebo> subbu: that impl class line I mean just like your method lookup example
<headius> we really only have AOT tidying and refinements left, right?
<enebo> subbu: but we do not want to ask that table if we can just store the value at the site
baroquebobcat has joined #jruby
<headius> enebo: if we did pre2 before cabin we'd have bug reports
<enebo> headius: yeah and this last kwargs work
<enebo> headius: which is less about figuring it out and more about repositioning it
<headius> ok
<headius> it would be supergreat to say we're supposed to be feature-complete for pre2
<enebo> headius: I sort of want the kwargs in interp and in something at beginning of method in JIT
<subbu> enebo, anyway, my crude off the cuff thoughts are that exposing repetitive runtime operations in IR => they can be optimized (unless they are so fine-grained that JVM can do that as well).
<headius> maybe it should be rc1
<enebo> subbu: ok. I guess I did not expect much more than that at this level
<headius> enebo: You mean an optimized kwargs pass-through? I want that too but not before pre2
<headius> we need to kick the next release out the door and start racing to the finish line
<enebo> subbu: If part of retrieving something is done n times in a scope perhaps we split those operations up so we can do that lookup once
<headius> very few people would be crazy enough to migrate straight to 9.0.0.0 anyway
<subbu> :)
<enebo> headius: no this is just so I can place pseudo-icky fix in two places vs n call locations
anaeem1 has quit [Ping timeout: 245 seconds]
<enebo> headius: but I need access in interp (and JIT) in required arguments
<enebo> headius: but block now uses Signature and method uses Arity and neither are viewable from IC
<enebo> headius: so a few things to tweak
<enebo> IC should get signature is this pass end goal
<enebo> then JIT and interp can have the dup + unfreeze logic added if kwargs method or closure
<enebo> headius: but we will need to seriously reconsider how we want to pass kwargs through internally
erikhatcher has joined #jruby
<enebo> headius: this both from perspective of simplicity of impl and also for direct wiring in those few cases where we do not need to really make a Ruby Hash
<headius> enebo: ohhh right, the pass through was because you need to defer some var logic later in stack
<headius> to fix those issues
<enebo> headius: but I also wonder if the MRI kwargs bugs you just mentioned a while ago will end up being a world changing event
<enebo> headius: seems like the further into the corners we go the more we discover :)
anaeem1_ has joined #jruby
camlow325 has quit [Remote host closed the connection]
camlow325 has joined #jruby
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:ef55159 by Chris Seaton): The build passed. (http://travis-ci.org/jruby/jruby/builds/54769323)
travis-ci has left #jruby [#jruby]
lucasallan has joined #jruby
calavera has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
erikhatcher has quit [Quit: erikhatcher]
anaeem1_ has quit [Ping timeout: 245 seconds]
bbrowning is now known as bbrowning_away
calavera has joined #jruby
iamjarvo has joined #jruby
nateberkopec has quit [Quit: Leaving...]
yfeldblu_ has joined #jruby
yfeldblum has quit [Read error: Connection reset by peer]
slyphon has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] chrisseaton pushed 1 new commit to master: http://git.io/hGoq
<JRubyGithub> jruby/master 4dceff3 Chris Seaton: [Truffle] Fix links in Rubinius core readme.
JRubyGithub has left #jruby [#jruby]
fivebats has quit [Quit: quit]
tcrawley-away is now known as tcrawley
mitchellhenke has quit [Quit: Computer has gone to sleep.]
kares has quit [Ping timeout: 245 seconds]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:a416467 by Kevin Menard): The build passed. (http://travis-ci.org/jruby/jruby/builds/54770032)
travis-ci has left #jruby [#jruby]
erikhatcher has joined #jruby
lucasallan has quit [Remote host closed the connection]
camlow325 has quit [Remote host closed the connection]
<asarih> headius: so… do you have Docker set up on your machine?
<asarih> headius: it's possible to run the container builds on it if you do.
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
nirvdrum has quit [Ping timeout: 246 seconds]
SynrG has quit [Ping timeout: 264 seconds]
camlow325 has joined #jruby
viking has quit [Quit: Leaving]
calavera has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
subbu is now known as subbu|busy
SynrG has joined #jruby
elia has joined #jruby
tcrawley is now known as tcrawley-away
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] bjfish opened pull request #2719: [Truffle] Adding back a Specialization used by fannkuch benchmark. (master...truffle_array_element_set_fix_fannkuch_again) http://git.io/hGHM
JRubyGithub has left #jruby [#jruby]
enebo has quit [Quit: enebo]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] chrisseaton pushed 2 new commits to master: http://git.io/hGQD
<JRubyGithub> jruby/master bd39f25 Brandon Fish: [Truffle] Adding back a Specialization used by fannkuch benchmark.
<JRubyGithub> jruby/master 6727b2c Chris Seaton: Merge pull request #2719 from bjfish/truffle_array_element_set_fix_fannkuch_again...
JRubyGithub has left #jruby [#jruby]
calavera has joined #jruby
calavera has quit [Max SendQ exceeded]
calavera has joined #jruby
tcrawley-away is now known as tcrawley
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
skade has joined #jruby
e_dub has quit [Quit: e_dub]
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
JRubyGithub has joined #jruby
<JRubyGithub> [jruby] chrisseaton fast-forwarded master from 6727b2c to 83f583d: http://git.io/hGF5
JRubyGithub has left #jruby [#jruby]
pjammer has joined #jruby
iamjarvo has joined #jruby
subbu|busy is now known as subbu
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:893aefe by Kevin Menard): The build passed. (http://travis-ci.org/jruby/jruby/builds/54771805)
travis-ci has left #jruby [#jruby]
lucasallan has joined #jruby
lucasallan has quit [Ping timeout: 265 seconds]
slyphon has joined #jruby
skade has quit [Quit: Computer has gone to sleep.]
jwinter_ has quit [Quit: Connection closed for inactivity]
travis-ci has joined #jruby
<travis-ci> jruby/jruby (master:acd4108 by Chris Seaton): The build passed. (http://travis-ci.org/jruby/jruby/builds/54773112)
travis-ci has left #jruby [#jruby]
baroquebobcat has quit [Quit: baroquebobcat]
baroquebobcat has joined #jruby
ModusPwnens has joined #jruby
<ModusPwnens> I've done a google search for this with no luck, but is it possible to use jruby in a rails initializer? I keep getting an unsupported major.minor version 52.0 when i try, even though the same code works fine in a rails controller
<lopex> ModusPwnens: rails under jruby ?
<ModusPwnens> yeah
<lopex> ModusPwnens: what java version do you have ?
<ModusPwnens> we just upgraded to java 8
<Antiarc> ModusPwnens: "unsupported major.minor version 52.0" usually means the wrong Java version
<lopex> ModusPwnens: make sure you dont have another java that is visible
<ModusPwnens> yeah i know, which is wierd, because I get that error if I try to create my object in the initializer, but I don't get it if I try to create it in my controller
<lopex> in $PATH and $JAVA_HOME
<ModusPwnens> it doesn't look like i do
<ModusPwnens> if I type where java
<ModusPwnens> i just get the one location that I know about
<lopex> and $JAVA_HOME ?
<lopex> or JDK_HOME, sometimes
<ModusPwnens> both java_home and jdk_home are empty
tcrawley is now known as tcrawley-away
<ModusPwnens> i know there is something odd happening though because i've seen this before if I'm not directly bootstrapped to this other workspace that only has java 8
subbu has quit [Quit: Ex-Chat]
DrShoggoth has quit [Quit: Leaving]
elia has quit [Quit: Computer has gone to sleep.]
baroquebobcat has quit [Quit: baroquebobcat]
baroquebobcat has joined #jruby
lucasallan has joined #jruby
baroquebobcat has quit [Client Quit]
bbrowning_away is now known as bbrowning
e_dub has joined #jruby
lucasallan has quit [Ping timeout: 256 seconds]
nirvdrum has joined #jruby