<kares>
tcrawley-away: the RG-proxy is working but seems terribly slow ... been downloading a 8M file locally for 3 minutes ... I still only got half ;(
yosafbridge has joined #jruby
yosafbridge has quit [Excess Flood]
yosafbridge has joined #jruby
yosafbridge has quit [Excess Flood]
pjammer has joined #jruby
yosafbridge has joined #jruby
yosafbridge has quit [Excess Flood]
yosafbridge has joined #jruby
skade has joined #jruby
yosafbridge has quit [Excess Flood]
yosafbridge has joined #jruby
benlovell has quit [Ping timeout: 240 seconds]
mister_solo has quit [Ping timeout: 276 seconds]
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
thsig has joined #jruby
benlovell has joined #jruby
mister_solo has joined #jruby
benlovell has quit [Ping timeout: 265 seconds]
cremes has quit [Remote host closed the connection]
<mkristian>
kares, yes, the build still hits the TB proxy. to avoid the TB I still need to upload the gems for the tests: rspec mainly. I started a while ago to get the build depending on maven-central only.
<tcrawley>
kares: mkristian: is there any way to determine the ip addr of the travis box? I can check to see if it has been blacklisted by the firewall rules
<mkristian>
I also made a ticket at travis GH to get the external IP to allow it for TB server.
<tcrawley>
ah
<mkristian>
yesterday I also asked on IRC but no answer
<tcrawley>
is travis on ec2?
<tcrawley>
kares: is the proxy still slow for you locally?
<mkristian>
I also made a ticket to ask them to add a rubygems proxy to their maven mirrors. that would help in general. can't tell you anything about travis really
mister_solo has joined #jruby
<mkristian>
tcrawley, the proxy sometime produces redirects, sometimes is slow, sometimes OK. but overall it feels normal to me - locally
<nirvdrum>
headius: I don't get any errors on Linux.
joast has joined #jruby
benlovell has quit [Ping timeout: 256 seconds]
enebo has joined #jruby
subbu has joined #jruby
Aethenelle has quit [Ping timeout: 272 seconds]
mister_solo has quit [Ping timeout: 265 seconds]
<kares>
tcrawley: mkristian much faster now locally ... thanks
<kares>
I believe travis-ci is on EC2 ... so those connections might be coming from it, right?
benlovell has joined #jruby
<ale_>
hello, if i start application from jruby script using Spoon.spawnp how can I grab the tty(?) so that I can send SIGINT through to the spawned process, currently I have to run two jruby processes and use Process.waitpid(Spoon.spawnp(...)) which reserves resources for no good reason because the first jruby process is waiting. any help is appreciated
mike___1234 has quit [Quit: Leaving]
yfeldblum has quit [Remote host closed the connection]
<headius>
nirvdrum: ok
<headius>
thanks
mister_solo has joined #jruby
<headius>
tcrawley, kares: travis is on EC2
<headius>
I'm not sure which region
<headius>
ale_: you may have better luck using the built-in spawn with JRuby 9k
<headius>
it uses real native process spawning and tty inheritance etc
<headius>
on JRuby 1.7 we were a bit more limited in how interactive subprocesses could be and tty was pretty hit or miss
<JRubyGithub>
[jruby] chrisseaton pushed 1 new commit to master: http://git.io/vknNG
<JRubyGithub>
jruby/master 454db3a Chris Seaton: [Truffle] Fix Exception#to_s
JRubyGithub has left #jruby [#jruby]
<chrisseaton>
headius: I've been trying to test the minitest in stdlib recently, and I don't think there's any tests for it - test/mri/minitest tests test/lib/minitest, not stdlib minitest, and they're different versions - does that sound crazy or is it quite possible that it's like that
<tcrawley>
kares: I didn't do anything to speed it up for you locally, but you are welcome :)
<tcrawley>
all I did was unblock the ec2 ip, let me know if tests are still failing with timeouts to the proxy
<kares>
I see ... lets check how travis-ci is doing than
<kares>
it's greenish now ... all should be fine, thanks again tcrawley :)
colinsurprenant_ has joined #jruby
colinsurprenant has quit [Ping timeout: 258 seconds]
colinsurprenant_ is now known as colinsurprenant
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] pinnamur opened issue #2998: Regular expression hang with string containing umlaut characters http://git.io/vknjd
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] eregon pushed 1 new commit to master: http://git.io/vkcez
<JRubyGithub>
jruby/master 61ad8cf Benoit Daloze: [Truffle] Fix handling of CoreMethod.constructor....
JRubyGithub has left #jruby [#jruby]
pjammer has joined #jruby
<kares>
mkristian: wouldn't the travis cache help not to put too much load on TB, do you know why it got disabled ?
<mkristian>
it got corrupted over time
<mkristian>
it looked like concurrency issues with the cache
<kares>
ah :( ... k, just asking - hopefully RG-proxy will run fine for a while
phlebas has quit [Ping timeout: 264 seconds]
<mkristian>
I think the better approach is to add the rubygems.org to there nexus instance which maven uses as mirror
<kares>
it's just scary how much jruby's projects now depends on it ...
<mkristian>
well, for jruby releases on master I will mirror the gems on maven-central
<mkristian>
but other projects . . . I do not know
<nirvdrum>
kares, mkristian: Now that jruby-openssl is out of the main repo, any thoughts on how I can set a breakpoint from within IDEA?
<kares>
nirvdrum: nope
<nirvdrum>
It's not mentioned in maven, so it doesn't appear on the classpath.
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] chrisseaton pushed 1 new commit to master: http://git.io/vkcUA
<JRubyGithub>
jruby/master f37dc99 Chris Seaton: [Truffle] Fix another Exception#to_s spec.
JRubyGithub has left #jruby [#jruby]
<mkristian>
I just send an email request to oss.sonatype.org and ask them to setup a proxy there.
<kares>
still, have not booted IDEA probably in years and counting :)
<nirvdrum>
I'll settle for other debug options ;-)
<kares>
mkristian: but that won't be the same as TB right? ... what kind of proxy you mean - a proxy proxying TB?
<mkristian>
nirvdrum, breakpoint inside BC ? just add BC to your classpath
<headius>
mkristian, kares: the maven cache is being repaired in travis but I don't know ETA
<mkristian>
kares, proxy to rubygems.org producing a maven repo
<mkristian>
nexus server has a plugin for it
<nirvdrum>
mkristian: I need to step into org/jruby/ext/openssl/X509Store.java and do so from a remote debug session.
<kares>
I see ... that's kind of a infra change ... I wonder if they do setup plugins based on user requests
<headius>
looks like I stubbed out most of it at some point
<mkristian>
kares, jruby-1.7.20 does not load the installed jruby-openssl and always uses the default gem from jruby. the difference to jruby-1.7.19 could be the different rubygems version coming with jruby
<headius>
ah...well jossl probably needs to be modified to support UNIXSocket (from jnr-unixsocket, which uses jnr-enxio)
<kares>
headius: nirvdrum: ... thought to give it a shot - seems like I might be able to get all from the UnixSocketChannel
<headius>
I question why someone's using ssl over a unix socket, though
<headius>
don't trust the kernel?
<nirvdrum>
Heh.
<kares>
headius: cause we can ... and we're Ruby-ists deal with it!
<headius>
damn you rubyists!
<lopex>
headius: equity
<kares>
mkristian: ok thanks ... will let it smoke a while hoping for next RG to fix it ... if not might try to change how we test
benlovell has joined #jruby
<mkristian>
kares, there will be fix for jruby-1.7.21 but 1.7.20 just does not work and will never work with any other jruby-openssl version as with the default gem
<headius>
kares, nirvdrum: doco online seems to indicate that this is a particularly weird thing to do
<mkristian>
kares, so you could skip this tests for 1.7.20
<mkristian>
rtyler, why not
benlovell has quit [Ping timeout: 256 seconds]
blaines has joined #jruby
<headius>
kares, nirvdrum: the twitterverse is having a good chuckle at the idea of SSL over a unix socket
<headius>
I guess it should work based on the Ruby ossl API, but it's weeeeird
<kares>
mkristian: I see ... did not run into that locally but now that you mentioned I checked and you are right
<kares>
quite annoying to have this bug in a few releases after it got "fixed"
<kares>
think it was behaving the same in 1.7.13 or 16 ;(
<mkristian>
kares, I thought there was some integration test for it. but did not look more.
benlovell has joined #jruby
<mkristian>
well, I am not sure if it is the same bug as before - let's see. there was another issue with 1.7.20 and mode 1.8 on those jossl test - will also look there
benlovell has quit [Ping timeout: 256 seconds]
jimbaker has joined #jruby
<lopex>
enebo: the algorithm might also be very inefficient since it goes like this
<lopex>
while traversing a tree, select ll nodes from that tree that yield true wrt that current node
<lopex>
*all
<lopex>
so it's already quadratic wrt tree size
<lopex>
(givent a predicate)
<enebo>
lopex: if your tree is unbounded sizewise then you will have problems regardless
<lopex>
what I managed to do it to cut the search once the node is found - that alone was not that obvious
<lopex>
I guess foretracking might solve the issue
<lopex>
enebo: it's like, for example for any many_to_one relation, traverse the tree, find original model definition and apply settings from original model to that that is being processed
jimbaker has quit [Ping timeout: 276 seconds]
<lopex>
enebo: or in lamers terms, given original model definition find where it's being used as dictionary and apply those settings to it
elia has joined #jruby
jimbaker has joined #jruby
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
robbyoconnor has quit [Quit: Konversation terminated!]
<Sinjo>
Having some trouble moving an app from MRI to JRuby 9k. Swapped out all the gems for JRuby compatible ones, but can't get jruby-pg to load. If I require "pg" in pry I get "LoadError: no such file to load -- pg_ext".
<Sinjo>
Kinda stumped on this one right now. Anyone know where I could start?
<headius>
Sinjo: pass -d, I suspect there's a problem loading the ext
<headius>
probably a changed API in 9k
<headius>
LoadError tends to mask those issues
<headius>
enebo: maybe we should allow non-Ruby exceptions during load to propagate out as-is
<Sinjo>
Does that want to go in JRUBY_OPTS?
<enebo>
headius: I admit it is really frustrating as a dev
<headius>
yeah
<headius>
enebo: it's not something we actively did...we just have a catch-all that re-wraps as an opaque LoadError
rsim has quit [Quit: Leaving.]
<Sinjo>
Nothing when I run the require, but some interesting stuff before that. Will drop it in a gist.
<JRubyGithub>
[jruby] headius pushed 1 new commit to jruby-1_7: http://git.io/vkWBG
<JRubyGithub>
jruby/jruby-1_7 a6db2cd Charles Oliver Nutter: Remove incorrect "eql" override that uses value equality....
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius pushed 1 new commit to master: http://git.io/vkWBu
<JRubyGithub>
jruby/master 7551aa7 Charles Oliver Nutter: Remove incorrect "eql" override that uses value equality....
JRubyGithub has left #jruby [#jruby]
<mjc_>
simi: activerecord-jdbc-adapter does this easily by just specifying the name, if you are just trying to connect to a DB with JNDI
sholden has joined #jruby
<sholden>
Hi everyone, I’m pretty new to Java but I’m working on migrating an older java api application to ruby. I’m trying to wire up a spring context based on the existing war file context xmls and dependency libs and running into plenty of exciting walls along the way.
<sholden>
The latest exception that I’m trying to understand is this:
<sholden>
SecurityError: sealing violation: package com.autoleadsystem.util is sealed
<sholden>
from org/jruby/javasupport/JavaUtilities.java:54:in `get_proxy_or_package_under_package'
<headius>
sholden: I don't even know what sealing is :-\
<sholden>
I can usually understand most java code, but I lack a lot of knowledge around the wider ecosystem and apis etc
<sholden>
I’m not sure why this is falling over when loaded from JRuby but not from the app itself in prod
<sholden>
lopex: yeah this application is pretty ancient
<headius>
sholden: another possibility is that we generate a utility class into the same package
<headius>
that may be more likely
<sholden>
sounds like a likely culprit
<headius>
perhaps there are interfaces from that package you try to implement in Ruby?
<headius>
or a class you extend
<headius>
why oh why don't you raise a SealedPackageException, Java
<headius>
at least we could handle this case specifically
<sholden>
so there are 2 different dependent jar files that both defining classes in that namespace but in different jars
<sholden>
but that doesn’t seem to be a problem in the existing setup
<sholden>
right now, I have a Beans class that is new’ing up a GenericApplicationContext and loading the bean definitions from the classpath and filesystem xml resources
<sholden>
when I do a context#refresh, that’s when this blows up
<headius>
huh...that article says they'd have to be all in the same jar
<headius>
I suspect this is partially inside the spring logic for wiring up JRuby
<headius>
they likely trigger an interface to be implemented to fulfill the injection, and that's where we go kablooey
<sholden>
If I require each of those 2 jars individually, then try and access a class from each, I get another exciting explosion. for science!
<sholden>
this jar gets required first: NameError: cannot link Java class com.autoleadsystem.util.PagUtil
<sholden>
and second: SecurityError: sealing violation: package com.autoleadsystem.util is sealed
<headius>
the first one is probably basically the same problem...it doesn't like the two jars getting loaded separately in the way we do it
<sholden>
cause by com.autoleadsystem.util.PagUtil.java_class and com.autoleadsystem.util.ClassUtil.java_class respectively
<headius>
the second jar seals the package so if you load it second it's ok
<headius>
or vice versa, whatever
<headius>
they are not consistently sealing
<simi>
mjc_, yup, but I want to load also another data from web container
<lopex>
sholden: is it an oracle/sap env related ?
<headius>
sholden: the inconsistent sealing we can't do anything about...perhaps when loaded into the container it just magically does them in the right order
<headius>
or if the error is happening very early in load, it's the same thing...wrong order loading those jars
<sholden>
no… this is usually run on a resin server and that wires up the app context? (i think… like I said, still kinda new to this ecosystem)
<headius>
the NameError usually indicates that you don't have the right jars loaded, but that seems odd if you managed to require both jars
<lopex>
maybe spring has a way to turn off that thing ?
<headius>
required jars go into the same classloader
<lopex>
headius: btw have you seen the previos conversation wrt that libgdx thingy ?
<sholden>
i reversed the require order, same result.
<lopex>
well week ago
<sholden>
not touchign the spring context at this point, just manually requiring those 2
<headius>
yeah, I've never liked that we wrap up some linkage and classloading errors as NameError
djbkd has quit [Remote host closed the connection]
<headius>
I think it comes down to the same problem: a single exception type that has specific causes we want to wrap and others we want to propagate
<lopex>
I was wrong about the NameError but yeah, still
<headius>
right
<headius>
anywhere we can isolate such a case we should stop wrapping
<headius>
since they're not recoverable anyway
<lopex>
that a good test case too
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
djbkd has joined #jruby
djbkd has quit [Read error: Connection reset by peer]
djbkd has joined #jruby
pjammer has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum opened issue #3001: [Truffle] Incorrect frame use for block variable access http://git.io/vkW93
JRubyGithub has left #jruby [#jruby]
mistergibson has quit [Quit: Leaving]
JRubyGithub has joined #jruby
JRubyGithub has left #jruby [#jruby]
<JRubyGithub>
[jruby] nirvdrum closed issue #3001: [Truffle] Incorrect frame used for block variable access http://git.io/vkW93
<Sinjo>
headius: Had a go at building jruby-pg locally. `rake build` produces "ext/java/org/jruby/pg/Connection.java:1434: error: method newEncoding in class RubyEncoding cannot be applied to given types; RubyEncoding rubyEncoding = RubyEncoding.newEncoding(context.runtime, encoding);"
<Sinjo>
headius: Likely 9k breakage?
<headius>
yeah could be
<headius>
it shouldn't be using that call
<headius>
it should be using runtime.getEncodingService().getRubyEncodingForEncoding or something like that
<headius>
not that any doco anywhere says that :-)
<Sinjo>
Ah right
<Sinjo>
Heheh
<Sinjo>
I'm gonna have to call it a night. Too much coding for one day. May have a go at a PR if I get some more time to spend on this.
<Sinjo>
Thanks for your help :)
subbu is now known as subbu|afk
<vikaton>
How fast is JRuby 9k ?
<headius>
unbelievably fast
<headius>
or not fast enough
<headius>
depends on what you're doing
subbu|afk is now known as subbu
<nirvdrum>
So, what exactly is a dvar? I'm having a bit of difficulty nailing that one down.
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<nirvdrum>
enebo: ^ whenever you're around.
<lopex>
nirvdrum: howdy there, I have absolutely no idea :)
<nirvdrum>
So it's because there's a block within a block?
<headius>
well, it's just because it is first assigned inside a block
<nirvdrum>
Ahh.
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<nirvdrum>
Well, that makes sense then.
<nirvdrum>
Thanks.
<headius>
all block-scoped vars are dvars
<headius>
method-scoped vars are lvars
<headius>
it's not a meaningful distinction
pjammer has joined #jruby
<headius>
block-local vars used to be looked up dynamically but that lookup became static at some point (probably 1.8)
<nirvdrum>
Yeah, I guess this one was weird because the var was defined in the method call, which usually is an lvar.
<lopex>
headius: meaning the evaluation is not consistent ?
<headius>
lopex: I imagine it wasn't before, but now they're both statically located the same way, via depth and offset
<lopex>
ok
<enebo>
hah…I still can see no reference to dvar in that link
<headius>
I remember enebo making that change in our parser WAY back in 2006 or so
<nirvdrum>
enebo: In that bug I linked to?
<enebo>
yeah
<lopex>
I hope block defaults etc are a bit consistent ?
<enebo>
yeah variables around 1.8.6 were lexical in nature but not implemented as lexical
<nirvdrum>
I didn't explicitly state dvar in it because I honestly wasn't sure if it was only dvars at the time. But I'm pretty sure it is. Truffle has a different issue with lvars in some cases that I need to deal with next.
<enebo>
we made them lexical in impl first
<headius>
so back to me...my patch above needs a touch of review, but it passes rubyspec and MRI tests
<headius>
enebo: ah, so they were actually made lexical for YARV
<headius>
that's right
<enebo>
headius: perhaps even 1.8.6 in a later Pl
<headius>
mmm
<enebo>
headius: ruby vars I think were always lexical
<nirvdrum>
headius: HTTP_PROXY is a bit funky. I'm not aware of any other JVM-based app using it. But I guess it makes sense.
<enebo>
headius: but the impl always did a lookup like they were not
<enebo>
headius: but always were lexically resolved so it was just a bad impl
<headius>
nirvdrum: that's just legacy stuff from MRI's logic
<headius>
I moved it down from find_proxy
<nirvdrum>
headius: As in, if you picked it up all of a sudden, it could confuse people that weren't expecting the JVM to pick it up.
<headius>
that's true, that's a difference from before
<enebo>
nirvdrum: sorry had you asked what is a DVarNode I would have figured this out sooner…I forgot we even named block vars that in AST
<nirvdrum>
It's how you configure proxies for most *nix apps, though. So if that's what MRI was doing, I could see people expecting it in that case.
<headius>
now that it's using the proper env mechanism it might pick up env vars rather than using JVM properties
<enebo>
nirvdrum: so yeah historical name still used by both impls
<headius>
nirvdrum: problem is I can't know which one they want in either case
<nirvdrum>
headius: I guess it'd be weird to have a proxy defined one way and expect it to be pointing to another proxy a different way. I've done that precisely once in my lifetime (for a weird use case Mogotest had).
<headius>
old logic ignored env
<nirvdrum>
headius: Yeah. Damned if you do, damned if you don't.
<headius>
new logic tries env first, and also honors noProxyHosts
<headius>
I think it's a tradeoff we have to make...if someone has both env and properties, they'll need to clear env
<nirvdrum>
Maybe give precedence to the JVM system properties and then fall back to the env var?
<headius>
hmm
<headius>
it is less likely to be set
<headius>
that might be ok
<nirvdrum>
Anyone currently setting it shouldn't see any difference in behavior that way.
<headius>
yeah I'll make that flip, that makes sense
<enebo>
headius: is line 27 same as MRI now?
<headius>
it is
<headius>
it's how we should have done this originally, but I didn't know about find_proxy
<enebo>
headius: ah ok so our patching was part of the issue?
tcrawley is now known as tcrawley-away
<enebo>
headius: since line 36 exists
skade has joined #jruby
<headius>
partially, yeah...the way we added it caused the JVM properties to force proxy use unconditionally, not honoring nonProxyHosts
<headius>
this way will honor that
<enebo>
headius: but how much of this patch could just be MRI lib version?
<headius>
what do you mean?
<enebo>
headius: or perhaps how much of it is in this patch?
<enebo>
headius: well only the ENV_JAVA parts should come from us right?
<headius>
the patch mostly just reverses our change on line 26 and makes their existing env logic also check JVM props
<enebo>
headius: the rest MRI should also have?
<headius>
right
<enebo>
ok
<headius>
I may submit a patch to them to break out the _from_env methods
<enebo>
headius: so >50% of this is just syncing up MRI again
mrmargol_ has quit [Remote host closed the connection]
<headius>
eregon: that refinement bug seems to be leaning the way we want... shugo agrees with me
<headius>
I'm sympathetic to _ko1's arguments but I don't think we can make that work in a consistent way
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
sholden has quit [Quit: sholden]
skade has quit [Quit: Computer has gone to sleep.]
vikaton has quit []
vikaton has joined #jruby
rsim has quit [Quit: Leaving.]
nateberkopec has quit [Quit: Leaving...]
Aethenelle has quit [Ping timeout: 244 seconds]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] chrisseaton pushed 1 new commit to master: http://git.io/vklgZ
<JRubyGithub>
jruby/master fa71f85 Chris Seaton: [Truffle] Tags in the wrong place.
JRubyGithub has left #jruby [#jruby]
<colinsurprenant_>
hiya!
baroquebobcat has quit [Quit: baroquebobcat]
<colinsurprenant_>
when a RubyHash or RubyArray is created, the inner objects are java.lang.String/java.lang.Long/… and not RubyString/RubyNumeric/… ?!
<colinsurprenant_>
for primitive types?
<chrisseaton>
Are you saying that's what you're seeing and it's a problem, or are you asking if that's how it works? It will be a RubyFixnum not a Long, and RubyString not a String
<chrisseaton>
(RubyString doesn't wrap a String anyway, it wraps a byte[] via ByteList)
<colinsurprenant_>
this is what I am seeing
djbkd has quit [Remote host closed the connection]
<colinsurprenant_>
(
<colinsurprenant_>
(yeah, I know about RubyString, thx)
deobalds has quit [Quit: Computer has gone to sleep.]
<chrisseaton>
Are you running a third party JRuby extension?
<colinsurprenant_>
hmmm… weird, where does that conversion takes place then
<colinsurprenant_>
my own extension I am writing
<chrisseaton>
oh right, well if you are creating the hash or array, you need to do it yourself
drbobbeaty has joined #jruby
KevinCorcoran has quit []
djbkd has joined #jruby
djbkd has quit [Remote host closed the connection]
elia has quit [Read error: Connection reset by peer]
<colinsurprenant_>
so, my ext wrapper defines a ruby method which takes a RubyHash as parameter, and passes that to my native Java object which accepts a Map parameter, then when I walk that Map, the inner object are java.lang.String and not RubyString
djbkd has joined #jruby
elia has joined #jruby
subbu has quit [Ping timeout: 250 seconds]
<chrisseaton>
Oh when you use the Java interfaces it automatically 'unboxes' to Java objects
<colinsurprenant_>
hmmmm
<colinsurprenant_>
by all mean, this is what I wanted, but I was expecting to have to do that manually :P
<mberg>
JRuby is reliably magic. It's almost disconcerting.
<colinsurprenant_>
not sure where that magic happen and this is puzzling me. my Map is Map<String, Object>, so I could understand the unboxing for the keys, but for the values, they are also unboxed …