<imperator>
lopex, well, what's happening is that i'm getting an invalid handle value returned from the GetCurrentProcess method on windows when using 64-bit jruby
<imperator>
so it doesn't segfault, it just errors out when i try to use that value
<imperator>
i tried hard coding it (to 0xFFFFFFFFFFFFFFFF) but then it complains about a bignum being too big to convert into a long
calavera has joined #jruby
<lopex>
ah
<imperator>
if there's some other type i can use to trick it, i'm happy to do so
diegoviola has joined #jruby
zeroecco has joined #jruby
<imperator>
hm, could use ulong_long i suppose
<lopex>
via ffi ?
<imperator>
yeah, just check for 'java' and bitness
<imperator>
if java and 64bit, then typedef a :handle to :ulong_long instead of :uintptr_t
<imperator>
clunky, but should work
<imperator>
now...if i can just rember how to tell if i'm using a 64 bit jruby...
<headius>
oh, it's some property I always have to look up
calavera has joined #jruby
<headius>
it's in ENV_JAVA somewhere
<Felystirra>
I'm not sure if this is the right place to post this question, but I'm asking it in consideration of my jruby app... When you're making a web request over ssl, are the URL and headers visible? or just the URL? or is the entire thing encrypted -- assuming no hijacks/breaks, the content and destination are encrypted?
<headius>
nothing at all should be visible
<headius>
the connection first negotiates encryption, and then you proceed from there as an HTTP request
<headius>
that's why russia had to block all of github rather than just the pages they didn't like...github is always ssl
<headius>
can't see what page was requested
<Felystirra>
so in theory, in that scenario, if someone in russia knew github's api ip, they could still use the api?
<headius>
I think they blocked the IPs
<headius>
they had to just block it completely
<imperator>
wut?
* imperator
missed this
<headius>
Russia blocked Github a couple times in the past month for having disagreeable content
<headius>
the latest one was a document listing the pros and cons of various methods of suicice
<imperator>
oh brother
<Felystirra>
oh, right. that did do an ip block vs a dns block.
<imperator>
the key thing here is that the path isn't found - in MRI that just returns nil
<imperator>
this a known issue or shall i report it?
iamjarvo has joined #jruby
iamjarvo has quit [Max SendQ exceeded]
iamjarvo has joined #jruby
iamjarvo has quit [Client Quit]
diegovio1 has joined #jruby
mister_s_ has quit [Ping timeout: 245 seconds]
diegoviola has quit [Quit: WeeChat 1.0.1]
calavera has quit [Ping timeout: 260 seconds]
calavera has joined #jruby
havenwood has joined #jruby
rcvalle has quit [Quit: rcvalle]
diegovio1 is now known as diegoviola
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] djberg96 opened issue #2291: Dir[] fails if path does not exist instead of returning nil http://git.io/fJ916w
JRubyGithub has left #jruby [#jruby]
tlarevo has quit [Ping timeout: 260 seconds]
yfeldblum has quit [Remote host closed the connection]
ruby-nuby has joined #jruby
<ruby-nuby>
hi. sorry if this is a stupid question, i am totally new to ruby. i need to consume 'maven' dependencies. i installed the jbundler gem, and i ran jbundle install. it created a .jbundle/classpath.rb file. but when i run my ruby code (in intellij) it does not pick up that classpath from that file.
<ruby-nuby>
it seems jbundler assumes i know how to use bundler, of which it is an extension for jruby i guess..
<ruby-nuby>
also, it seems to build the JBUNDLER_CLASSPATH wrong
<ruby-nuby>
it keeps adding Jars.home and then ':' - which is not correct for windows
<ruby-nuby>
so the classpath it builds is not usuable.
<ruby-nuby>
even after i fix it, if i do require 'java' and require 'jbundler' - i still get class def not found errors
dcolebatch has quit [Ping timeout: 258 seconds]
brettporter has quit [Remote host closed the connection]
<asarih>
nirvdrum: time-dependent tests are bad… just saying.
<nirvdrum>
asarih: Not disagreeing. On failed yesterday because the seconds rolled over from 2 to 3.
<nirvdrum>
*One failed
vtunka has joined #jruby
<asarih>
nirvdrum: wheeeee
<nirvdrum>
asarih: What are these run on anyway? m1.small?
<enebo>
asarih: but for knowing whether blocking is blocking forever you sort of need some time limit to fail vs hang forever
<asarih>
nirvdrum: it's a Blue Box instance. 1.5 VCPU, 3 GB memory.
<enebo>
I am not saying time limits are great but for blocknig tests how else do you fail?
<asarih>
enebo: I agree with you!
<nirvdrum>
asarih: Ahh. I thought you guys were on AWS.
<asarih>
nirvdrum: some are. but they are not EC2 instances.
Aethenelle_ has joined #jruby
<asarih>
nirvdrum: i.e., Docker containers running on large EC2 instances
Aethenelle has quit [Ping timeout: 245 seconds]
Aethenelle_ is now known as Aethenelle
<nirvdrum>
asarih: Ahh. I was having a hard time reconciling that in my head.
<enebo>
nirvdrum: You can commit this to a longer time value I don’t know what it is doing but it cannot be testing <1s as any thing more than it blocked forever. Aethenelle can change it later if we are wrong
<Aethenelle>
?
<nirvdrum>
enebo: I didn't look into it. I'm not sure if it semantically is supposed to be < 1s or was just an arbitrarily chosen time.
<enebo>
Aethenelle: test ffi function on travis is hitting >1s on an ffi spec
triple_b has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<enebo>
nirvdrum: no not really. I doubt this was a microopt either (sometimes people would do stuff like this so it did not need the JIT to inline it).
<enebo>
nirvdrum: You can change it if you want
<nirvdrum>
Cool.
<nirvdrum>
The inlining bit was why I was asking.
<nirvdrum>
I'm just coming across some stuff as I go along here.
<claudiuinberlin>
guys, what king of jvm profiler with jruby should I use?
<enebo>
nirvdrum: If we do that we should end up commenting on it in code
<nirvdrum>
Agreed.
skade has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/wNiEQg
<JRubyGithub>
jruby/master 238637a Kevin Menard: There seems to be no point in manually inlining a method here.
JRubyGithub has left #jruby [#jruby]
mister_s_ has quit [Ping timeout: 260 seconds]
diegoviola has quit [Ping timeout: 260 seconds]
sferik_ has joined #jruby
sferik has quit [Read error: Connection reset by peer]
e_dub has quit [Quit: e_dub]
diegoviola has joined #jruby
claudiuinberlin has quit [Read error: Connection reset by peer]
claudiuinberlin has joined #jruby
mister_s_ has joined #jruby
e_dub has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] enebo pushed 1 new commit to jruby-1_7: http://git.io/JV37Dg
<JRubyGithub>
jruby/jruby-1_7 084d55e Thomas E. Enebo: Bump for new version
<Aethenelle>
there an easy way to handle an #ifdef in one of these .ffi files?
diegoviola has joined #jruby
shellac_ has quit [Quit: Computer has gone to sleep.]
mkristian has joined #jruby
skade has joined #jruby
skade has quit [Client Quit]
<Aethenelle>
huh.... we could use the rbx etc files if we wanted... not sure we really do as a blanket policy since they create platform dependant code...
<Aethenelle>
we don't seem to be using many of these generated files actually... lib/ruby/stdlib/ffi/platform/**/*.rb
x1337807x has joined #jruby
zeroecco has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
lanceball is now known as lance|afk
<enebo>
lopex: (foo)+?
<Aethenelle>
though, they may have been largely suplanted by jnr-constants and jnr-posix
phrinx has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] djberg96 opened issue #2293: uintptr_t is wrong size on 64-bit JRuby http://git.io/RWYn2w
JRubyGithub has left #jruby [#jruby]
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
brettporter has joined #jruby
e_dub has quit [Quit: e_dub]
x1337807x has joined #jruby
iamjarvo has joined #jruby
benlovell has joined #jruby
x1337807x has quit [Client Quit]
x1337807x has joined #jruby
brettporter has quit [Ping timeout: 258 seconds]
benlovell has quit [Ping timeout: 264 seconds]
<Aethenelle>
SQUEE!! getting some very nice info and ideas for creating qemu VMs
<Aethenelle>
should have a working sparc-solaris machine soon enough...
dumdedum has quit [Quit: foo]
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
josh-k_ has quit [Remote host closed the connection]
<Aethenelle>
question: how does the amazon hosting get paid for?
<enebo>
Aethenelle: Who are you asking?
<Aethenelle>
I'm trying to figure out hosting for the various VMs I'll need to set up for FFI... was wondering if i could piggyback on jruby's arrangement w/ amazon (assuming one existed)...
<enebo>
Aethenelle: we don’t have one. We are using a hosted instance from engineyard but we will be trying to move off that soon since we only do static pages for www.jruby.org on it now
<Aethenelle>
the actual download is from S3... engineyard taking care of that too then?
<enebo>
Aethenelle: they are at least currently
<enebo>
Aethenelle: for S3 I am not sure if we will be moving those to someone elses account or not
<Aethenelle>
k... I'll keep looking around then...
<enebo>
Aethenelle: k
<rtyler>
1.7.17
<rtyler>
YAYYYYYY
<rtyler>
WOOOOOOO
<enebo>
rtyler: Can you set milestones and such? I don’t remember?
<enebo>
on issues
<Aethenelle>
hrm... maybe i can talk w/ the distros I use about hosting them for vagrant...
robbyoconnor has quit [Ping timeout: 256 seconds]
<enebo>
Aethenelle: It would so great to be able to hook up to an instance of each platform we support
<Aethenelle>
is there an official list somewhere?
<Aethenelle>
pretty sure AIX and maybe a couple others are going to have to be hardware still
<enebo>
Aethenelle: I just look at jnr-ffi Platform enum
<Aethenelle>
k
<enebo>
Aethenelle: which does not include versions of architectures of those OSes
<enebo>
os390 might also be hard
<enebo>
:)
<Aethenelle>
enebo: yes, not gonna happen... maybe.. at least not w/ vagrant
zeroecco has joined #jruby
<rtyler>
enebo: yes, I can now, what's up
<enebo>
People who enjoy fixing jruby bugs…please consider 1.7.18 targetted issues. They should be that hard and are things we should get fixed sooner than later
<rtyler>
I'll see what we can do
<enebo>
rtyler: remember to set milestone on issues and also perhaps let mkristian to know again
<Aethenelle>
huh... there's a couple 390 emulators...
* rtyler
nods
<rtyler>
now I have to write up an internal announcement
<rtyler>
a chore I'll gladly do :D
<enebo>
very few people ever mark milestones and I spend time each release doing it. I am sure I sometimes get the milestone wrong :)
<rtyler>
for jruby-gradle it's the only sane way I can generate the changelogs
<Aethenelle>
i've been going through the jira tickets recently myself
<mkristian>
:)
<enebo>
I just realized right now that I have not been doing it for PRs :)
<enebo>
mkristian: heya!
<mkristian>
enebo, hi
<enebo>
mkristian: I did not change to -SNAPSHOT manually so you have a virgin landscape
<enebo>
mkristian: I sent an email about rmvn not working due to I think cyclical boostrapping issue
<enebo>
mkristian: or maybe just ordering perhaps
<mkristian>
enebo, cool, that should be working but it does not - let's see
<enebo>
(I did not realize mkristian was on or I would not have asked rtyer to remind him)
<mkristian>
to put labels - yes I got the message
yfeldblum has joined #jruby
<rtyler>
enebo: "wrong encoding of “hen string interpolation is involved"
<rtyler>
I think that's supposed to say 'is resolved'
<enebo>
rtyler: perhaps fixed at front is less confusing?
lance|afk is now known as lanceball
<rtyler>
that works too
<enebo>
we had an issue where encoding was wrong when interpolation was involved
<nirvdrum>
enebo: "wrong encoding of “hen string interpolation is involved" ?
<enebo>
I will change it on the website though
<nirvdrum>
Heh. rtyler was similarly confused.
<nirvdrum>
Wow. The popen stuff got fixed. I missed that.
<enebo>
- fixed wrong encoding of `` contents when string interpolation is involved
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<enebo>
nirvdrum: is that better?
<enebo>
nirvdrum: if not give me the proper sentence :)
<nirvdrum>
enebo: So, interpolation within backticks?
<Aethenelle>
there's a few IBM people flowing in and out of here... maybe we can con them out of a license/installer
<enebo>
nirvdrum: yeah
<enebo>
Aethenelle: yeah I would think it would be in their best interests
<Aethenelle>
123notit
<enebo>
Aethenelle: It is not like we will be making money off of a licensed instance or anything
x1337807x has joined #jruby
<enebo>
whoever smelt it dealt it
<Aethenelle>
whoever denied it supplied it...
<nirvdrum>
enebo: I'd just leave the symbols out and say "backticks" to lower the likelihood of being misread. In a proportional font, it can be a bit hard.
<enebo>
mpapis: geez I guess I forgot to tell you 1.7.17 is out :|
<nirvdrum>
enebo: Although if you do make another change, I guess I'd make the "Fixed" lowercase for consistency. But I'm trying to not let it bother me :-P
<enebo>
hahah
<enebo>
nirvdrum: I am leaving it in just because I know that fact now :P
<nirvdrum>
This may shock you, but I'm rather anal about that stuff.
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] elyscape opened issue #2294: String#crypt fails on Windows http://git.io/HClKig
<cpuguy83>
I guess the builder is trying to pull stuff that doesn’t exist anymore?
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] jeremyevans opened issue #2295: Regexp.union [] gives wrong result in ruby 1.8 mode http://git.io/yilZAw
JRubyGithub has left #jruby [#jruby]
<chrisseaton>
cpuguy83: correct - this commit was relying on a snapshot release, which will always be changing
<jeremyevans_>
enebo: headius: while the bug I just opened isn't a security vulnerability in jruby, it could potentially cause a vulnerability in applications running jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] phrinx opened issue #2296: NullPointerException with JRuby 1.7.17 running rspec & simplecov in debug mode http://git.io/AD6mSg
JRubyGithub has left #jruby [#jruby]
erikhatcher has joined #jruby
diegovio1 has joined #jruby
diegoviola is now known as Guest95927
diegovio1 is now known as diegoviola
abedra has joined #jruby
Guest95927 has quit [Ping timeout: 252 seconds]
<abedra>
headius: I should have reached out earlier, but you are more than welcome to swing by the Groupon office any time today. We've got plenty of space for you.
mister_s_ has quit [Ping timeout: 258 seconds]
abedra has quit [Ping timeout: 250 seconds]
abedra has joined #jruby
<enebo>
jeremyevans_: I see. I guess based on number of remaining 1.8.7 JRuby users (I am guessing is getting pretty small) plus needing to accept [] we can probably wait and get this for 1.7.18
<enebo>
marked against 18
brettporter has joined #jruby
<chrisseaton>
Named captures in a regexp - where are these set as locals? I can't any code relating to them at all.
<enebo>
chrisseaton: I would guess ParserSupport
x1337807x has joined #jruby
<chrisseaton>
enebo: I mean after the match is made, at runtime, where are the captured strings copied into local variables - that's not going to be in ParserSupport is it?
<enebo>
chrisseaton: ah yeah. no it will not be in parsersupport only defining the lvars is there
<chrisseaton>
enebo: I still can't see that though - 'capture' doesn't find it for me - what could I search for
mje113__ has joined #jruby
<chrisseaton>
enebo: ah sorry found it in allocateNamedLocals - just need to find where the values are set now...
<enebo>
chrisseaton: yeah I also need to find it. I thought I had added the feature as well :) unless I only hooked up parser
<chrisseaton>
enebo: you pass the relevant spec
<enebo>
chrisseaton: but that is only alloc’ing the lvars in static scop
<enebo>
chrisseaton: something must know that the named patterns are actually locals on the other side (e.g. execution of regexp)
bbrowning_away is now known as bbrowning
<chrisseaton>
enebo: yeah - and I can't find any reference to it anywhere
<enebo>
chrisseaton: you find real software using this?
<chrisseaton>
enebo: no - I didn't know they were an unusual feature though? are they new? we're just going through the last of the RubySpec language specs and this was one
<enebo>
chrisseaton: Added in 1.9.3? I think but they are not very common
<enebo>
chrisseaton: ordering of regexp vs string you are matching on and also needing to be a literal makes them more uncommon
<chrisseaton>
enebo: ok - thanks for helping me find that - I was looking in RubyRegexp etc and didn't think to look in Helpers
<enebo>
chrisseaton: I do use them but I always get the order wrong “” =~ // vs // =~ “”
josh-k_ has joined #jruby
<enebo>
chrisseaton: how does Truffle++ back/implement their DirectedGraph impl?
<enebo>
Adjacency lists?
<chrisseaton>
enebo: you mean how is the Graal IR represented?
<enebo>
chrisseaton: well I am assuming their is a directed graph in there somewhere at some point
<enebo>
there
<enebo>
chrisseaton: I am only asking because I have the most memory innefficient DG on the planet and I am putting it on a diet
<chrisseaton>
You can see two children are inline, so that's an optimisation
josh-k has quit [Ping timeout: 250 seconds]
renderful has joined #jruby
renderful has quit [Remote host closed the connection]
<chrisseaton>
But funny you should ask this question today! We're at Google Munich this week speaking to Google compiler people, and V8 are working on a new IR, which is sea of nodes like Graal, and they were talking about how a node is composed of two parts...
iamjarvo has joined #jruby
skade has joined #jruby
<chrisseaton>
They have the graph part that is small, just references to parents and children, and it refers to a more complicated node object, which is shared where possible, and contains more info
iamjarvo has quit [Client Quit]
<enebo>
chrisseaton: I might be more interested specifically in com.oracle.graal.graph.Graph but I can look at it
<enebo>
chrisseaton: right now my first pass wil just be to make in and out lists on vertex share a single primitive array vs 2 HashSets which probably will be enough of a diet for the moment
<enebo>
oh I see
<chrisseaton>
why are they hash sets? is each edge labelled? we rely on an order convetion
nateberkopec has quit [Quit: Leaving...]
<chrisseaton>
the thing with the V8 is if you have nodes which often do the same thing, like call #+ and the arity is one, you can share that info between nodes
<enebo>
chrisseaton: well I think we only wanted the set behavior
<enebo>
chrisseaton: just in case our CFG algo I think accidentally added the same edge twice
<enebo>
chrisseaton: we label edges
<enebo>
chrisseaton: which is somewhat unrelated but we could have two edges between same two vertices but still exist because they have different labels
<chrisseaton>
that's ok with an ordering convention
nateberkopec has joined #jruby
<enebo>
I would have to study CFG and DFG algos to know whether we rely on order of edges…I did not start with a background on compilers :)
<enebo>
but if we do I should probably add some specs to make sure order is preserved