<chrisseaton>
But that doesn't explain the class path being different?
<chrisseaton>
sorry that last bit wasn't a question
<sebastianedwards>
yeah, i just tried removing jruby-truffle.jar as part of the docker build.. same boottime issue, so i guess it wasn't that flag at all
<chrisseaton>
how are you using JRuby? Binary tarball? Building from source?
<sebastianedwards>
running puma in source dir
<chrisseaton>
sebastianedwards: ok this is a bit beyond my knowledge, can't really help any more, sorry
<sebastianedwards>
same issue when using rake tasks or console
<chrisseaton>
sebastianedwards: try tomorrow on here central time
<sebastianedwards>
jirb boots quick.. just goes to hell when requiring rails environment
<sebastianedwards>
chrisseaton: Will do. Thanks
r0bby_ has joined #jruby
mrmargolis has joined #jruby
lopex has quit [Quit: Connection closed for inactivity]
dinfuehr has joined #jruby
dinfuehr has quit [Ping timeout: 256 seconds]
<sebastianedwards>
Interesting.. running a profile and it looks like SecureRandom may be the culprit
<Antiarc>
sebastianedwards: Might be low entropy on your production box?
<sebastianedwards>
Antiarc: Unsure, if I boot jirb and call SecureRandom.hex the first call will take a couple minutes.. subsequent calls return instantly. Would that be consistant with low entropy?
<Antiarc>
Possibly - if it's trying to prefill an entropy pool before doing the first random ops. Try running an entropy daemon and see if that helps.
<Antiarc>
egd is what I normally use, but it may be different based on your platform
<Antiarc>
That'll tell you how much entropy is in your system entropy pool. If that's too low, then cryptographic operations will block until there is enough entropy available.
<sebastianedwards>
what range would be too low?
<Antiarc>
I don't quite recall, but like under 200, I think?
<sebastianedwards>
Yeah, I have ~150
<Antiarc>
Yeah, see if egd or rng-tools helps
<Antiarc>
May just be that you're entropy-starved
colinsurprenant has quit [Quit: colinsurprenant]
<Antiarc>
you should only use rng-tools if your hardware supports it, otherwise you'll reduce the randomness of your cryptographically-used entropy
<sebastianedwards>
Antiarc: Thanks for your help. Will try egd and see how it goes :)
<Antiarc>
FWIW, the "normal" way of adding more entropy is to do a lot of IO
r0bby_ is now known as robbyoconnor
pjammer has joined #jruby
colinsurprenant has joined #jruby
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
mrmargolis has quit [Remote host closed the connection]
colinsurprenant has quit [Quit: colinsurprenant]
Hobogrammer_ has quit [Ping timeout: 256 seconds]
mrmargolis has joined #jruby
mrmargolis has quit [Remote host closed the connection]
dinfuehr has joined #jruby
Hobogrammer has joined #jruby
x1337807x has joined #jruby
Hobogrammer_ has joined #jruby
dinfuehr has quit [Ping timeout: 276 seconds]
Hobogrammer has quit [Ping timeout: 276 seconds]
Hobogrammer has joined #jruby
Hobogrammer_ has quit [Ping timeout: 246 seconds]
Hobogrammer_ has joined #jruby
Hobogrammer has quit [Ping timeout: 250 seconds]
deobalds has joined #jruby
Hobogrammer_ has quit [Ping timeout: 272 seconds]
deobalds has quit [Quit: Computer has gone to sleep.]
Hobogrammer has joined #jruby
nateberkopec has quit [Quit: Leaving...]
Hobogrammer has quit [Ping timeout: 240 seconds]
colinsurprenant has joined #jruby
Hobogrammer has joined #jruby
dinfuehr has joined #jruby
dinfuehr has quit [Ping timeout: 255 seconds]
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
colinsurprenant has quit [Quit: colinsurprenant]
subbu has joined #jruby
subbu has quit [Client Quit]
mrmargolis has joined #jruby
bjfish2 has quit [Quit: bjfish2]
bjfish2 has joined #jruby
deobalds has joined #jruby
kfpratt has joined #jruby
mrmargolis has quit [Remote host closed the connection]
dinfuehr has joined #jruby
dinfuehr has quit [Ping timeout: 250 seconds]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] eregon pushed 1 new commit to master: http://git.io/vJaxr
<JRubyGithub>
jruby/master 43e6382 Benoit Daloze: [Truffle] Remove fixed TODO
JRubyGithub has left #jruby [#jruby]
bjfish2 has quit [Quit: bjfish2]
vikaton has quit [Quit: Connection closed for inactivity]
kfpratt has quit [Remote host closed the connection]
dinfuehr has joined #jruby
dinfuehr has quit [Ping timeout: 256 seconds]
aemadrid has quit [Quit: Connection closed for inactivity]
tenderlove has quit [Remote host closed the connection]
tenderlove has joined #jruby
drbobbeaty has joined #jruby
ponga has joined #jruby
deobalds has quit [Quit: Computer has gone to sleep.]
deobalds has joined #jruby
dinfuehr has joined #jruby
dinfuehr has quit [Ping timeout: 264 seconds]
dinfuehr has joined #jruby
dinfuehr has quit [Ping timeout: 272 seconds]
sebastianedwards has quit [Remote host closed the connection]
dinfuehr has joined #jruby
havenwood has quit [Remote host closed the connection]
arturaz has joined #jruby
drbobbeaty has quit [Read error: Connection reset by peer]
justinmcp has quit [Ping timeout: 256 seconds]
dinfuehr has quit [Read error: Connection reset by peer]
drbobbeaty has joined #jruby
dinfuehr has joined #jruby
justinmcp has joined #jruby
deobalds has quit [Quit: Computer has gone to sleep.]
mister_solo has joined #jruby
tenderlove has quit [Read error: Connection timed out]
tenderlove has joined #jruby
shellac has joined #jruby
rsim has joined #jruby
mister_solo has quit [Ping timeout: 264 seconds]
digitalextremist has quit [Ping timeout: 256 seconds]
ponga has quit [Remote host closed the connection]
skade has joined #jruby
drbobbeaty has quit [Ping timeout: 240 seconds]
marr has joined #jruby
vtunka has joined #jruby
mister_solo has joined #jruby
digitalextremist has joined #jruby
vtunka has quit [Quit: Leaving]
deobalds has joined #jruby
shellac has quit [Quit: Computer has gone to sleep.]
digitalextremist has quit [Ping timeout: 252 seconds]
ponga has joined #jruby
shellac has joined #jruby
ponga has quit [Read error: No route to host]
ponga has joined #jruby
ponga has quit [Read error: Connection reset by peer]
erikhatcher has joined #jruby
skade has quit [Quit: Computer has gone to sleep.]
lopex has joined #jruby
erikhatcher has quit [Quit: erikhatcher]
shellac has quit [Quit: Computer has gone to sleep.]
drbobbeaty has joined #jruby
yh__ has joined #jruby
tcrawley-away is now known as tcrawley
shellac has joined #jruby
ponga has joined #jruby
vtunka has joined #jruby
skade has joined #jruby
ponga has quit [Quit: Leaving...]
shellac has quit [Quit: Computer has gone to sleep.]
ponga has joined #jruby
ponga has quit [Quit: Leaving...]
yfeldblum has quit [Ping timeout: 265 seconds]
pjammer has joined #jruby
vtunka has quit [Quit: Leaving]
skade has quit [Quit: Computer has gone to sleep.]
vtunka has joined #jruby
bbrowning has joined #jruby
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
nirvdrum has joined #jruby
pjammer has joined #jruby
bbrowning has quit [Quit: Leaving]
bbrowning has joined #jruby
skade has joined #jruby
bjfish2 has joined #jruby
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
vikaton has joined #jruby
nateberkopec has joined #jruby
brightball has quit [Quit: Leaving...]
<headius>
bjfish2: looks like it connects but doesn't load
lance|afk is now known as lanceball
<headius>
bjfish2: it did eventually come back for me but it's very slow
colinsurprenant has joined #jruby
colinsurprenant has quit [Client Quit]
<vikaton>
how fast is 9k pre2 ?
<headius>
so fast you won't believe it
<headius>
or not as fast as you'd hope
<headius>
vikaton: depends what you're doing, but generally 9k should be as fast or faster than 1.7
<headius>
we're seeing many cases be faster just due to having a smarter compiler
<vikaton>
oo nice
pierreatkillbill has joined #jruby
ponga has joined #jruby
<headius>
I'd say give it a shot and let us know :-)
KevinCorcoran has joined #jruby
KevinCorcoran has quit [Ping timeout: 264 seconds]
<nirvdrum>
headius: Did you see that match3 issue I filed?
<nirvdrum>
Thanks. I need to get better at navigating those things.
<nirvdrum>
Yeah, that'd be match2.
<headius>
sorry, test/mri/ruby/test_regexp
<nirvdrum>
There's a similar test in test_string.rb
<nirvdrum>
And it looks like we currently exclude it.
deobalds has quit [Quit: Computer has gone to sleep.]
dumdedum has joined #jruby
<headius>
hooray, +1 for MRI tests
<nirvdrum>
Yeah. Sorry for not finding that earlier.
<headius>
nirvdrum: no worries...this is why we have both
shellac has quit [Quit: Computer has gone to sleep.]
pierreatkillbill has joined #jruby
insidious has joined #jruby
deobalds has joined #jruby
pjammer has joined #jruby
rsim has quit [Quit: Leaving.]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/vJrlP
<JRubyGithub>
jruby/master bf7414b Kevin Menard: [Truffle] Tagged failing grpowned? specs.
JRubyGithub has left #jruby [#jruby]
<headius>
hmm, no enebo yet
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
rsim has joined #jruby
gregorsc5 has joined #jruby
KevinCorcoran has joined #jruby
gregorsc5 has quit [Client Quit]
<chrisseaton>
nirvdrum: so what's the story with grpowned? or do you just know those ones are failing?
<nirvdrum>
We don't even have it implemented.
<nirvdrum>
I don't know how it got untagged in the first place.
tenderlove has quit [Remote host closed the connection]
KevinCorcoran has quit [Ping timeout: 240 seconds]
tenderlove has joined #jruby
<chrisseaton>
Bulk untagging I think - maybe we should stick to bulk untagging classes that we're actively working on
<chrisseaton>
Although I suppose that wouldn't have stopped this
tenderlove has quit [Client Quit]
gregorsc5 has joined #jruby
<nirvdrum>
I take that back. I was calling it wrong. Instead there's a mess of internal errors. But I'm still not sure if we've really tried to implement this.
<headius>
I'm playing with reducing them down to one instr
<headius>
Match2 and Match3 were almost identical
<enebo>
ok
brightball has joined #jruby
<enebo>
headius: are you fixing with a extra boolean or some fix like that?
<chrisseaton>
Are we not able to come up with more descriptive names than Match, Match2, Match3 etc
<enebo>
chrisseaton: well they come from MRI
<chrisseaton>
oh ok
<enebo>
chrisseaton: but I agree in spirit
TheBrayn has joined #jruby
<TheBrayn>
hi
<enebo>
chrisseaton: It was useful for AST comparisons back at the time but we can certainly fix the names in IR. I think we just chose the same names … for not particular reason
<nirvdrum>
enebo: I asked headius earlier. But maybe you have some insights as to why the receiver and value nodes seem to be reversed in match3?
<nirvdrum>
We didn't properly reverse them back, so Truffle has been running the wrong =~ for a little while now :-/
<enebo>
nirvdrum: match3node?
subbu has joined #jruby
<nirvdrum>
Yeah. receiverNode seems to always be the regexp literal, but the receiver should be the value node. Unless I'm terribly mistaken.
<nirvdrum>
IR seems to reverse them back.
<enebo>
nirvdrum: Let me look at grammar…I think this may be the // =~ “” vs “” =~ // thing
<enebo>
nirvdrum: yeah match2 is // =~ which supports named-captures potentially and match3 is “” =~ //
vtunka has quit [Quit: Leaving]
<nirvdrum>
enebo: I eventually worked that out. But I think the receiverNode and valueNode references in Match3Node should be logically swapped. Or maybe just renamed.
<enebo>
nirvdrum: I think there appears to also be some other tiny behavioral difference besides captures but order does not really matter in either case
<enebo>
nirvdrum: oh so second, first is converted back to first, second in IR?
<nirvdrum>
Well, in your example, "" is the receiver.
<nirvdrum>
But in the Match3Node, // is the receiverNode and "" is the valueNode.
vtunka has joined #jruby
<enebo>
But all these are calling op_match or variant where regexp is the receiver no?
oko has left #jruby ["peace ! i'm out !!"]
<nirvdrum>
No. "" =~ // call String#=~
<enebo>
nirvdrum: This might have been a arbitrary choice and history
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] chrisseaton pushed 1 new commit to truffle-head: http://git.io/vJrM0
<JRubyGithub>
jruby/truffle-head fc94c95 Chris Seaton: Merge branch 'master' into truffle-head
JRubyGithub has left #jruby [#jruby]
shellac has joined #jruby
<enebo>
nirvdrum: in 1.8 =~ was not a method and this was hard-coded so it would have been regexp as receiver
<headius>
they're probably reversed because that's the order the come out of the parser
<headius>
and in MRI AST nodes are just dumb structs
<enebo>
nirvdrum: but in 1.9 it is still regexp as reciever if sttring ant not-overridden
<headius>
we don't need the abstraction because we can just reverse it right on the way in
arturaz has quit [Ping timeout: 246 seconds]
<enebo>
nirvdrum: but I think the 1.8 is the reason
<enebo>
headius: this is not from that though
<headius>
no?
<enebo>
headius: well no the parser swaps the order for match3 to make regexp the receiver
<nirvdrum>
enebo: That makes sense. Would it be terribly disruptive to reverse? I could just rename the fields to avoid changes to the parser itself.
<enebo>
headius: I think this is because in 1.8 regexp woul also be the receiver for match3
<enebo>
headius: 1.9 was first version you could override =~
<headius>
I'm confused...do all these just call =~ on the regexp or not
<enebo>
headius: you are working on the bug where it is not
<enebo>
headius: is type override =~ explicitly it calls that and not regexp match
<headius>
I'm working on relevant changes to match building
<headius>
wait, what?
<enebo>
Isn’t that what nirvdrum reported?
<headius>
"str" =~ /exp/ should dyncall if Regexp has a different =~
<headius>
but it always calls against regexp, no?
<enebo>
notice he is redfining it on string?
<headius>
oh hmm
<nirvdrum>
It should call it against String.
<headius>
yeah, I'm getting match2 and match3 reversed in my head
<enebo>
headius: match3 is weird because str comes first
<nirvdrum>
Sorry for any confusion there.
<nirvdrum>
The issue I filed against IR is you skip the dispatch and just call the String =~ op.
<headius>
match2 is always // =~ "" and will dyncall against regexp
<enebo>
nirvdrum: you can swap it on the AST node so long as you fix the reordering in IRBuilder
<enebo>
nirvdrum: I think that might make this less confusing
<enebo>
headius: yeah match2 // is lhs
<headius>
and not while I'm working on the same code :-)
<enebo>
:)
<headius>
so why can't match2 just dyncall =~ on LHS every time
<headius>
er, match3, dammit
<enebo>
headius: nirvdrum: let’s just agree receiver is regexp for match2 and str for match3
<nirvdrum>
enebo: Cool. I just wanted to make sure a) there wasn't some meaning here I was getting wrong; and b) that it wouldn't require deeper parser massaging.
<headius>
sure
<headius>
the optimization we had in place was that if str instanceof String directly call op_match19 on regexp
<enebo>
nirvdrum: you uncovered more ancient history I think :)
<headius>
my fixes for this make all matches the same, all just dispatch =~ to LHS
<nirvdrum>
headius: I think you should always dyncall on LHS for match3.
<enebo>
headius: yeah probably just a bypass to avoid calling through str
<headius>
where LHS is $_ in case of MatchInstr
<enebo>
headius: perhaps the days of microopting
<nirvdrum>
Clearer names in general would be helpful, too.
<headius>
nirvdrum: I can't think of a behavioral reason not to always do the same in match2 and match
<nirvdrum>
But this one was downright lying to me ;-)
<enebo>
so match_str and match_regexp maybe is better names for 3 and 2 but what is 1 again?
balo has quit [Ping timeout: 272 seconds]
<headius>
enebo: so far tests are passing with all of these just becoming calls to =~
<enebo>
oh match_implicit
<headius>
yeah
<nirvdrum>
Technically it could be any BasicObject in match3.
<enebo>
match_str, match_regexp, an match_implicit
pjammer has joined #jruby
<headius>
~/foo/ I think
<enebo>
headius: good renames for instrs?
<nirvdrum>
=~ is defined in Kernel.
<headius>
enebo: it wil lall just be MatchInstr
<enebo>
headius: do you still want these callapeds
<headius>
< CallBase
<nirvdrum>
And in fact, that's how mspec matchers work and was the root of the problem I was hitting.
<headius>
nirvdrum: I just had specs break before I fixed the ordering :-)
<headius>
same case
<nirvdrum>
bjfish2: I'm getting false where all other implementations are returning true.
<enebo>
headius: oh so since these may be dyndispatch you want it to be calls
<headius>
but all seems green now with single MatchInstr that always dyncalls
<headius>
enebo: yeah
<headius>
they'll just work like any other call
<headius>
in fact I'm not sure we need MatchInstr now
<headius>
the capture logic happens in separate instrs
<enebo>
headius: so you will emit %t_1 = gvar($_) ?
<headius>
yes
<enebo>
headius: ok
<headius>
and yeah I don't even need MatchNode now
<headius>
oh hmm
<headius>
$~ detection
<bjfish2>
nirvdrum works on OS X and travis, what’s your OS again ?
<nirvdrum>
headius: I think there might be a difference in scoping for captured vars in match2 vs match3. I'd verify that before collapsing to a single handler.
<nirvdrum>
bjfish2: Linux.
<enebo>
headius: It might be worth keeping it as its own operation type for $~ and still extend callbase
<headius>
match2 does scoping but IR builder builds that logic separate from the match
<headius>
I mean does named captures in scope
<enebo>
yeah there is a special node type for in it AST
<nirvdrum>
bjfish2: Specifically, LinuxMint 17.1 (an Ubuntu 14.04 derivative) with an updated 3.16 kernel. I'm testing on a directory using ecryptfs.
<enebo>
which IR does not take advantage of
<headius>
no reason to really
<headius>
we have the match in hand
<nirvdrum>
I'd point at ecryptfs as the discriminator, but MRI, JRuby, and Rubinius all return the correct answer.
<headius>
I will post a gist for review momentarily
deobalds has quit [Quit: Computer has gone to sleep.]
skade has quit [Quit: Computer has gone to sleep.]
<enebo>
headius: sorry IR does take advantage of the capturenode…I realize that was what you meant by emitting separately
iamjarvo has joined #jruby
insidious has quit [Ping timeout: 256 seconds]
insidious has joined #jruby
camlow325 has joined #jruby
<nirvdrum>
headius: Incidentally, I think we had that anonymous module name search wrong because our specs were inadvertently passing due to our incorrect handling of match3.
<nirvdrum>
It's fun how all this stuff spirals out of control.
<headius>
ain't it fun building a language runtime?
<headius>
I'm going to file this as a bug with MRI anyway because this is a nasty little hidden perf issue
shellac has quit [Quit: Computer has gone to sleep.]
jph98 has quit [Quit: jph98]
tenderlove has joined #jruby
<nirvdrum>
Yeah. Since we were reversing the callers, it just called Regexp#=~ and then fell out of the spec block.
<nirvdrum>
Surprisingly, we only had 8 false positives that way.
<headius>
nice
dinfuehr has quit [Remote host closed the connection]
donV has quit [Quit: donV]
KevinCorcoran has joined #jruby
mister_solo has quit [Ping timeout: 240 seconds]
pierreatkillbill has quit [Ping timeout: 264 seconds]
<nirvdrum>
bjfish2: I'll try to see what's going on shortly.
havenwood has joined #jruby
KevinCorcoran has quit [Remote host closed the connection]
<keeguon>
Hey, we’re trying to compile code using warbler but we’re running into an issuez where we think we have a file exceeding the 65536 limit in Java since it’s jRuby classes loads the Ruby code as is in the Java class and puts it into a variable. Is it possible to specify an option at compile time to either skip all the comments in a Ruby file to gain some lines or maybe reading the file rather than completely loading the
<keeguon>
content in the Java file which may or may not avoid this limit (not a Java expert)?
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius closed issue #2748: RubyGems of JRuby1.7.19 is v2.4.5 (should be v2.1.9) http://git.io/hpBi
JRubyGithub has left #jruby [#jruby]
donV has joined #jruby
keeguon has quit [Quit: keeguon]
<bjfish2>
nirvdrum maybe this isn’t ready yet then if Process.groups isn’t ready
<headius>
enebo: I'm going to do a pass over jnr project issues/PRs and get them released
<nirvdrum>
I see what the problem is.
<headius>
we kinda want to release them to get the tidied up licensing and IP anyway
<nirvdrum>
I'm just surprised this wasn't an issue for you guys.
<enebo>
headius: yeah. s sounds great
skade has joined #jruby
keeguon has joined #jruby
<headius>
ugh, why do people come in and ask a question and then a minute later
<nirvdrum>
bjfish2: Out of curiosity, what values do you get back from Process.groups?
<headius>
keeguon: hey, you're back
<keeguon>
Sorry got disconnected
<headius>
keeguon: that's a bug
<headius>
I guess I never considered that case
lanceball is now known as lance|afk
<chrisseaton>
nirvdrum: I'll take a look in a bit
<keeguon>
headius: do you want me to file an issue on gh?
<chrisseaton>
nirvdrum: I remember Process.groups is a funny one - no actual jar-posix call It hink
shellac has quit [Quit: Computer has gone to sleep.]
<nirvdrum>
chrisseaton: I have a fix.
<nirvdrum>
chrisseaton: A silly issue Java is supposed to shield you from ;-) The primitive node is offsetting by ARRAY_LONG_INDEX_SCALE rather than ARRAY_INT_INDEX_SCALE. So every other entry is being skipped and then we're writing beyond the end of the array.
<chrisseaton>
That's why they call it Unsafe!
<nirvdrum>
I'll push the fix momentarily.
<nirvdrum>
I was just curious what you guys were seeing for Process.groups.
zorak8 has quit [Quit: Leaving]
deobalds has joined #jruby
pjammer has quit [Ping timeout: 265 seconds]
<keeguon>
headius: I’m surprised it hasn’t come up before, I guess large Ruby files aren’t that common
<headius>
keeguon: yeah me too
<headius>
file a bug, it won't be a difficult fix
<enebo>
keeguon: on 9k?
<enebo>
keeguon: we have an open bug on a too large error already I thin
<enebo>
keeguon: actually file away. This one may not be due to too large of a body
<headius>
master should work ok for normal jrubyc
<headius>
I added splitting
<headius>
jrubyc --java would have the same problem on 9k and 1.7 I think
<headius>
that embeds full source as a string
<enebo>
headius: yeah I know I just had someone file a bug last week because I thought it was a size issue with chunking
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/vJoSf
<JRubyGithub>
jruby/master 49de4a5 Kevin Menard: [Truffle] Properly index into the Rubinius::FFI::Platform::POSIX#getgroups array.
JRubyGithub has left #jruby [#jruby]
<enebo>
headius: it appears to bve something else though
<nirvdrum>
Okay. So you're seeing the skipping around as well.
<nirvdrum>
chrisseaton: How about bin/jruby -X+T -e 'p Process.gid' ?
<chrisseaton>
20
<nirvdrum>
And there we have it.
<chrisseaton>
All those zeros?
<nirvdrum>
Since we were offsetting by the length of a long in an int array, they're just the default zeroed out values.
<nirvdrum>
Your user is part of 13 groups. The first N/2 showed up in the array. The second half are written past the array (not sure what Unsafe does here).
<chrisseaton>
Just writes into the heap - can cause a VM crash
<nirvdrum>
In your case, your Process.gid happens to fall in that first half. In my case, it fell into the second half.
<nirvdrum>
So it was enough to give you guys a pass where I was seeing a failure.
<nirvdrum>
Anyway, fixed now. And we understand the problem.
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius pushed 1 new commit to master: http://git.io/vJoNq
<JRubyGithub>
jruby/master 8b1ddba Charles Oliver Nutter: Use latest snapshots of changing jnr projects.
JRubyGithub has left #jruby [#jruby]
<headius>
we'll see how these go
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius pushed 1 new commit to jruby-1_7: http://git.io/vJoNW
<JRubyGithub>
jruby/jruby-1_7 c0ca176 Charles Oliver Nutter: Use latest snapshots of changing jnr projects.
JRubyGithub has left #jruby [#jruby]
<headius>
enebo: if you're fixing that issue, your build will be the right one to see that jnr updates don't break anything
<headius>
I'm grabbing lunch while these settle
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/vJoxr
<JRubyGithub>
jruby/master 63cf6d3 Kevin Menard: [Truffle] Untagged more passing File specs.
JRubyGithub has left #jruby [#jruby]
pjammer has joined #jruby
balo has joined #jruby
donV has quit [Quit: donV]
x1337807x has joined #jruby
x1337807_ has joined #jruby
x1337807_ has quit [Max SendQ exceeded]
x1337807x has quit [Read error: Connection reset by peer]
x1337807x has joined #jruby
yh has joined #jruby
x1337807x has quit [Ping timeout: 240 seconds]
aramisbear has joined #jruby
x1337807x has joined #jruby
insidious has quit [Ping timeout: 264 seconds]
<enebo>
headius: ok. no known build yet since I don’t have any idea what broke
brightball has quit [Ping timeout: 252 seconds]
<headius>
recursive namespace fix is looking ok so far
<headius>
test-extended still running but mostly through the big items
<enebo>
yeah it just made it to test_command_line.
<nirvdrum>
enebo, headius: If not much of a hassle, I was hoping to get a new repo under the JRuby org called "jruby-truffle-stack". It'll house small apps and libs that we're trying to get running in JRuby+Truffle, particularly before they're baked enough to land in the project proper.
<headius>
I pushed updates to jnr snapshots on master and 1.7, once those look good I'll release the lot and bump to release versions
<headius>
nirvdrum: I have no problem with that
<enebo>
nirvdrum: yeah and we can always remove once you guys get further a long
<nirvdrum>
At your leisure then, please.
<headius>
if you feel it's worth putting under the org, I'll spin it
<nirvdrum>
headius: Eh? If you need to kill a build, go for it.
<nirvdrum>
I'll catch any failures later in the day.
<headius>
it won't really delay catching failures, just expand the possible changesets that might have caused them
<fuzzyhorns1>
anyone use syntastic in vim while doing jruby development?
<headius>
have not ever used that before
dinfuehr has joined #jruby
<headius>
is this a leading question? :-)
<fuzzyhorns1>
i like using syntastic in general for ruby dev, but one sad thing is when im working in jruby it spins up the jvm on every syntax/style check
<fuzzyhorns1>
which is very slooowww
<fuzzyhorns1>
so trying to see if anyone else found a way around that c:
<headius>
I figured that's where you were going
<headius>
I'd recommend configuring it to just use MRI always
<headius>
there must be a way
shellac has joined #jruby
<headius>
we could try to hack around it with flags and background JVMs but the simplest answer for you is to just wire MRI directly
<fuzzyhorns1>
yeah and i thought i could do that via syntastic's settings (just point to mri ruby bin) but it doesnt seem to be working
<fuzzyhorns1>
ah well, something to complain to scrooloose about :d
<headius>
yup yup, good luck
<headius>
some day maybe we'll be able to compete on startup time :-)
dinfuehr has quit [Ping timeout: 264 seconds]
<headius>
I don't know why I'm staring at these builds...practically nothing changed in the jnr projects
<headius>
I guess ppc64 stuff is big but I'm not going to see tests for that
erikhatcher has quit [Quit: erikhatcher]
<headius>
asarih: when can we expect PPC64LE support in travis? :-D
<asarih>
headius: does it need hardware?
<enebo>
asarih: and PDP11?
<headius>
I'm sure there's emus out there
<enebo>
JRuby on a PDP11…nope
<headius>
enebo: when we start getting JRuby issues, yes :-)
<headius>
JRubMe
<asarih>
Hardware, not any time soon. Emulator, it's up to the users, I'd guess.
donV has joined #jruby
deobalds has quit [Quit: Computer has gone to sleep.]
<enebo>
PDP-11 is a series of 16-bit minicomputers
<headius>
asarih: I'd love to see us spin up emulators for all our non-x86 platforms in JNR
<headius>
test across a dozen platforms under travis
<enebo>
that would be a tight fit memory wise :) probably could not get our Fixnum cache in that
<headius>
ahh 16-bit
<headius>
I bet the 64-bit floating point math just screams on that thing
<enebo>
I always think of PDP-11 because it is mentioned in yacc and Jay source code
<enebo>
That code is opt’d for PDP-11
<headius>
mruby would probably work
<enebo>
It is fucking amazing this claims PDP11 sold into the 90s
<enebo>
It came out in the 70s
<headius>
I was trying to get mruby to compile for an embedded processor I was playing with
<headius>
I think I blew the code segment cap though
deobalds has joined #jruby
<headius>
enebo: motorola chips from late 70s clung to life well into the 90s
<enebo>
pretty funy to see the size differences of releases of these things
<enebo>
goes from a big board stuffed with chips to a single tiny chip with space for pin outs
<enebo>
maybe Moore’s law is boring for most, but it still entertains me
<headius>
I assume pdp-11 evolved a bit over twenty years
<headius>
like C64s ran off a 6510, beefier version of 6502 from early game consoles
<headius>
imagine how our current rates of resource utilization would look to a 6502 programmer from 1979
<headius>
it's probably half waste
<enebo>
“There were other significant innovations in the Q-Bus lineup. For example, a system variant of the PDP-11/03 introduced full system Power-On Self-Test (POST)."
<enebo>
hmmm why isn’t this fucker booting…don’t know grab me another board
<headius>
enebo: jnr updates look fine on master, I'm going to go ahead with releases
<enebo>
headius: awesome
dyce_ is now known as dyce
<enebo>
headius: I will like celebrating by going to Firkin's
<headius>
lovely day for it
<headius>
we're hoping to get over to Sidhe this week
<headius>
new brewery on east side
marr has quit [Ping timeout: 244 seconds]
deobalds has quit [Quit: Computer has gone to sleep.]
mje113__ has joined #jruby
<headius>
enebo: have you had maven release plugin not update pom.xml right?
<headius>
I don't recall changing anything but it's putting the tag "1.2.9" on a snapshot revision and not pushing a non-snapshot commit
<enebo>
headius: I do not use release plugin ever
<headius>
hmm
<headius>
I don't know your method
<enebo>
headius: mostly because when we were testing using it on maven changeover to jruby the particular version of Java and git were horrendously unreliable
<enebo>
I manutally update the version and git stuff and then use maven deploy -Poss-sonatype-something-else-in-this-string
<headius>
ah
<headius>
well that's easy enough
<enebo>
headius: if you search on maven release and git you will see tons of people wtf’ing about how flaky it can be
<enebo>
headius: which you would expect to be fixed eventually?
<headius>
it doesn't push tags either
<headius>
it just pushes the pom change
<enebo>
headius: well…it might :) on some magic combo
<headius>
I always end up with some unpushed crap afterwards
<headius>
and some pushed
<enebo>
headius: yeah it is pretty weird
<nirvdrum>
enebo: Can you add the "JRuby core" team to that repo when you get a chance?
pierreatkillbill has joined #jruby
<enebo>
nirvdrum: ok
<enebo>
nirvdrum: done
<nirvdrum>
Thanks again.
skade has quit [Read error: Connection reset by peer]
skade has joined #jruby
<bjfish2>
nirvdrum what’s the mew repo all about?
<nirvdrum>
bjfish2: I just pushed to it.
<bjfish2>
i saw that now thanks
<nirvdrum>
This is that repo I told you had I been working on but needed to clear before opening. It's all set now.
<nirvdrum>
roda/ is the small web app I was trying to get going.
<nirvdrum>
It runs a basic test through rack-test currently.
shellac has quit [Quit: Computer has gone to sleep.]
<nirvdrum>
It could still do with some tidying up, I think. But it's good enough to get started with.
<nirvdrum>
The general idea is use JRuby to install dependencies locally with bundler, then switch to Truffle to load them up.
mkristian has joined #jruby
<nirvdrum>
And the truffle_workarounds.rb is where we stash project-specific shims until we get things working JRuby+Truffle, proper.
fuzzyhorns1 has quit [Read error: Connection reset by peer]
fuzzyhorns1 has joined #jruby
<mkristian>
enebo, headius the test-maven-3.3.1 is green now and I could merge it to master. BUT it means that build hacking needs mvn-3.3.1 or one needs to use ./mvnw
<mkristian>
dito for building any distribution artifacts under ./maven directory
<headius>
mkristian: ok, thanks for the update
<enebo>
mkristian: I think I am cool with this for master…this means no more pom.xml merging when we merge from jruby-1_&?
<enebo>
1_7
<headius>
remaining concerns, I guess, are places where we can't rely on 3.3.1 having any problems with the build
<headius>
like linux dists
<enebo>
mkristian: I am hoping we can gitignore the poms
<headius>
they won't just use mvn 3.3.1 because we demand it
<enebo>
good question. I wonder how fast Java tooling gets pushed out on most distros
<enebo>
headius: although they can still build with mvnw but perhaps they don
<enebo>
’t like that idea much
<mkristian>
you can build jruby with any mvn at least 'mvn clean install' the local build
<chrisseaton>
bjfish2: you are welcome to contribute to the repo gems that you might want to see if you can get running
<chrisseaton>
and the shims become a nice todo list, and the whole thing can be used for testing
<mkristian>
but for jruby-complete and the gem and the tar.gz, etc you need mvn-3.3.1
subbu is now known as subbu|lunch
yfeldblum has joined #jruby
<mkristian>
enebo, under ./maven there will be no more pom.xml files but for ./pom.xml, lib/pom.xml, core/pom.xml and truffle/pom.xml I thought to keep them - for people who can not use mvn-3.3.1
<mkristian>
and IDEs
shellac has joined #jruby
<mkristian>
what about jenkins ? it might fail with the change
<enebo>
mkristian: ok. I am happy that 1.7 merges will end up being easier if we can .gitignore all the others then it will only be there few files
mrmargolis has quit [Ping timeout: 256 seconds]
skade has quit [Quit: Computer has gone to sleep.]
<mkristian>
enebo, and it helps if we all use the same "version" of maven ruby-dsl via maven-3.3.1 - helps reproducing problems ;)
shellac has quit [Quit: Computer has gone to sleep.]
<enebo>
mkristian: well I have less a problem with the pom.rb usage than killing off pom.xml for jruby-1_7 since it is a stable branch for other disters
<headius>
1.7 doesn't use pom.rb does it?
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<enebo>
headius: we generate pom.xml from them don’t we?
<headius>
I don't think so
<headius>
not on 1.7
<headius>
only on master
<enebo>
but we have pom.rb files in jruby-1_7?
<mkristian>
headius, already on 1_7 the pom.xml are generated for some time now
<headius>
really?
<headius>
there's no core/pom.rb, what am I missing?
<mkristian>
yes, to reduce the merge problems. yes not everywhere. but under ./maven
<headius>
ahh, I guess I did not know that
<headius>
I only updated pom.xml because that was the only file I saw
pjammer has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<headius>
ok, so how do any of our merge problems improve?
<headius>
we're still merging pom.rb, maybe it's easier
<headius>
I assume pom.rb would need to live in root and core if we removed 1.7 pom.xml's
<headius>
in which case they're a merge issue
<headius>
right?
<mkristian>
yes, for core/ lib/ and root there will be merge issues. basically redoing the changes inside the respective pom.rb - I can put the pom.rb on 1_7 as well !?
mrmargolis has joined #jruby
<mkristian>
I do not think to many people really do changes on those pom.xml/pom.rb files.
drbobbeaty has quit [Read error: Connection reset by peer]
<headius>
mkristian: there may be small things we can do to make them merge nicer, but current policy of just re-doing changes isn't awful
subbu is now known as subbu|away
<headius>
as long as we do it on a timely basis
drbobbeaty has joined #jruby
<headius>
1.7 poms mostly just change to update versions now anyway, which they track independently
<chrisseaton>
headius enebo I put together a nifty way of benchmarking without having to show people graphs and talk about mandelbrot - http://i.imgur.com/DW4Gpmb.png - it's in that stack repo
<chrisseaton>
it also brilliantly demonstrates warmup
<headius>
nice, heatmaps
<headius>
now what does it mean? :-)
<chrisseaton>
it's just an old-school demo scene plasma effect
<mkristian>
headius, I not keen to changes any build things on 1_7 but also not really oppose it either. getting rid of pom.xml on master will help
<chrisseaton>
each implementation sends pixel data down a pipe and this application renders it and reports fps
<headius>
mkristian: yeah I don't have any real interest in twiddling 1.7 build
<headius>
I was just trying to understand what merge might look like
mrmargol_ has joined #jruby
mrmargolis has quit [Ping timeout: 245 seconds]
rsim has quit [Quit: Leaving.]
<mkristian>
headius, what about jenkins ? it might fail with the change. is there a way to install mvn-3.3.1 on jenkins
mrmargol_ has quit [Remote host closed the connection]
<headius>
hmm, I'm not sure
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius pushed 1 new commit to jruby-1_7: http://git.io/vJKjv
<JRubyGithub>
jruby/jruby-1_7 01e7076 Charles Oliver Nutter: Update to latest jnr releases.
JRubyGithub has left #jruby [#jruby]
<headius>
enebo: jnr stuff is current on both branches
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius pushed 1 new commit to master: http://git.io/vJKjt
<JRubyGithub>
jruby/master f196dbb Charles Oliver Nutter: Update to latest jnr releases.
JRubyGithub has left #jruby [#jruby]
skade has joined #jruby
subbu|away is now known as subbu
pjammer has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius closed issue #2790: Jruby to use upcoming jnr-ffi release to have ppc64le TypeAliases http://git.io/jFVf
<JRubyGithub>
[jruby] headius closed issue #2373: Redirecting $stdout to an object fails with: Errno::EBADF: Bad file descriptor - Bad file descriptor http://git.io/2j5ChA
JRubyGithub has left #jruby [#jruby]
_djbkd has joined #jruby
ponga has quit [Quit: Leaving...]
Liothen has quit [Ping timeout: 246 seconds]
lanceball is now known as lance|afk
Liothen has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius reopened issue #2373: Redirecting $stdout to an object fails with: Errno::EBADF: Bad file descriptor - Bad file descriptor http://git.io/2j5ChA
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius closed issue #2373: Redirecting $stdout to an object fails with: Errno::EBADF: Bad file descriptor - Bad file descriptor http://git.io/2j5ChA
JRubyGithub has left #jruby [#jruby]
dinfuehr has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius closed issue #2368: pipes/handles are closed when the jruby main thread exits, which can lead to unexpected behavior http://git.io/Ly87TQ
JRubyGithub has left #jruby [#jruby]
<headius>
chrisseaton, nirvdrum, bjfish2: btw, when you merge a PR please tag it with the next JRuby release it will go in
<headius>
we have a 100% milestone policy in the issue tracker
<headius>
for closed items
fuzzyhorns1 has left #jruby [#jruby]
<chrisseaton>
headius: there's a truffle-dev milestone we use - are there some issues or PRs without it?
<headius>
there's a bunch of PRs with no milestone at all
<enebo>
I had to handle exactly 50 issues last week
<headius>
looks like 52 that mention truffle
<enebo>
only some were truffle
x1337807x has joined #jruby
<headius>
I know I miss some sometimes
<enebo>
headius: wait you say 52 right now?
<chrisseaton>
ok I'll take a look
<enebo>
headius: I just went through all for PR1 and resolved all unmilestoned issues last week
<chrisseaton>
yeah i see them
<headius>
enebo: I'm reviewing unmarked now because 1.7.20 is issue free
<headius>
it's PERFECT now
drbobbeaty has quit [Read error: Connection reset by peer]
<enebo>
headius: ah ok. I was like wow someone closed a lot of issues recently :)
<headius>
PRs mostly
<enebo>
headius: ah I have not handled PRs though
drbobbeaty has joined #jruby
<enebo>
headius: the script I made did not look at PRs but I think it was written before they exposed those as issuable? If that makes sense
<headius>
right
<headius>
probably before they made them separate
<chrisseaton>
is there a search term I can use to find things without milestones? milestone:none isn't it
<headius>
if your script doesn't see them they must be a different api?
<headius>
chrisseaton: milestone dropdown in table header
<headius>
there's one for "no milestone"
<enebo>
chrisseaton: under mileston there is drop down
<headius>
oh, it's "no:milestone"
<headius>
weird
yfeldblum has quit [Remote host closed the connection]
<chrisseaton>
that's literally backward
<headius>
it is
<headius>
I never would have guessed that
<headius>
search with a separate field for things that aren't there
<headius>
so with the truffle ones tidied up, there's 94 unmarked items, 87 of them PRs
<headius>
going back years
<headius>
I guess we haven't audited this in a while
<chrisseaton>
about six months ago I asked in a lot of them if the issue still applied
<chrisseaton>
so a lot of them are probably ripe to close with no-further-feedback
<headius>
I'm talking about closed stuff
<headius>
that has no milestone marked
<chrisseaton>
oh sorry
<headius>
that probably wasn't clear before...I meant that everything that's closed needs to have some milestone set
<headius>
just to keep things tidy
<enebo>
headius: just mark them all against invalid or something
<enebo>
headius: I mean perhaps review the newest page or two
<headius>
enebo: might as well...we're not going to correlate
<headius>
PRs usually have an issue attached that we DID mark, too
<enebo>
headius: yeah people can resolve if they really want by correlating commit to tag
dinfuehr has quit [Remote host closed the connection]
<chrisseaton>
headius: sorry I'm just making a mess here - I'll leave you to it
<chrisseaton>
all the Truffle ones are done though
subbu is now known as subbu|afk
dinfuehr has joined #jruby
dinfuehr has quit [Remote host closed the connection]
lance|afk is now known as lanceball
yfeldblu_ has joined #jruby
<headius>
chrisseaton: yeah no problem
<headius>
enebo: I went back to roughly 1.7.19 timeframe and marked appropriately
<headius>
anything merged before that I just punted into invalid
<headius>
(of PRs)
<headius>
I marked all issues correctly
<enebo>
headius: great
yfeldblum has quit [Ping timeout: 265 seconds]
KevinCorcoran has quit [Ping timeout: 240 seconds]
tcrawley is now known as tcrawley-away
subbu|afk has quit [Ping timeout: 244 seconds]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] chrisseaton pushed 1 new commit to truffle-head: http://git.io/vJ6wM
<JRubyGithub>
jruby/truffle-head 4ab6c57 Chris Seaton: Merge branch 'master' into truffle-head
JRubyGithub has left #jruby [#jruby]
<nirvdrum>
headius: Are there any special steps to building jnr-posix? Or is it just "mvn package"?
pierreatkillbill has quit [Ping timeout: 246 seconds]
<enebo>
nirvdrum: that should work
<enebo>
headius: huzzah test_dash_0 is back
<nirvdrum>
enebo: Thanks. I see a bunch of failures on Windows that don't seem to be masked out. Do you just skip tests when packaging?
<nirvdrum>
Or do I have something wrong here?
<enebo>
nirvdrum: yeah that is expected and unfortunate behavior
<enebo>
nirvdrum: on windows I -DskipTests=true because I suck
<enebo>
nirvdrum: I will say I tried solving one of them and then got frustrated when it did not work
<nirvdrum>
That's fine. The README is just scant and there's no wiki. I didn't know how crazy this all got.
<enebo>
nirvdrum: fix it
<enebo>
:)
<nirvdrum>
Will do.
<enebo>
nirvdrum: you are my hero
<nirvdrum>
No one's ever said that to me before!
<enebo>
nirvdrum: get used to it hero
<nirvdrum>
Unless they're doing the Cameron voice.
<enebo>
nirvdrum: are you trynig to do some truffle stuff for windows?
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<nirvdrum>
It's not a huge priority. But I'd like to poke at it.
<enebo>
nirvdrum: ok. You wlil realize we have some jruby test issues as well
<enebo>
nirvdrum: but I suppose if you fix jnr-posix issues some of it will start working
<enebo>
nirvdrum: OTOH you wlll also find people do not think about wnidows when they write tests
<nirvdrum>
Some of these methods look like they're implemented trivially in Windows. I'm hoping they're not working just because no one took the time to do so.
<enebo>
nirvdrum: some for sure
<enebo>
nirvdrum: others seem like they should work because MSDN say they will but then you realize how awful MS libc support is
bbrowning is now known as bbrowning_away
<enebo>
nirvdrum: I feel bad whenever I think about windows because I don’t do enough to support it
<nirvdrum>
Others still seem like the could be implemented in Java straightforward enough. E.g., unlink.
<enebo>
nirvdrum: but there is only so much time and maybe not enough loud folks to complain
<enebo>
nirvdrum: unlink is fd or char*?
<nirvdrum>
char* I believe.
<enebo>
yeah…
<enebo>
nirvdrum: int _wunlink( const wchar_t *filename );
<nirvdrum>
Yeah, the wchar stuff is what I'm not clear on.
<enebo>
nirvdrum: you will find out when native is supported you should use wchar_t variants
<nirvdrum>
Windows supports both, generally. But I don't know what comes down from FFI.
<enebo>
nirvdrum: there are other w* methods in WindowsLibc
<nirvdrum>
I've really never worked with FFI before.
<enebo>
nirvdrum: copy those pattern-wise
skade has quit [Read error: Connection reset by peer]
<enebo>
nirvdrum: I think the only weird thing about wchar_t is it is terminated with \0\0