<JRubyGithub>
[jruby] chrisseaton pushed 1 new commit to truffle-head: http://git.io/x87tgQ
<JRubyGithub>
jruby/truffle-head 672834f Chris Seaton: Merge branch 'master' into truffle-head
JRubyGithub has left #jruby [#jruby]
benlovell has joined #jruby
mbj has quit [Quit: leaving]
shellac has joined #jruby
mkristian has joined #jruby
mrmargolis has joined #jruby
colinsurprenant has joined #jruby
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Hobogrammer has joined #jruby
Hobogrammer_ has joined #jruby
skade has quit [Quit: Computer has gone to sleep.]
Hobogrammer has quit [Ping timeout: 264 seconds]
lanceball is now known as lance|afk
skade has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] eregon pushed 1 new commit to master: http://git.io/T5hQJg
<JRubyGithub>
jruby/master cf6c3c5 Benoit Daloze: [Truffle] Fix a few obvious FindBugs bugs.
JRubyGithub has left #jruby [#jruby]
benlovel1 has joined #jruby
benlovell has quit [Quit: Lost terminal]
johnmuhl has quit [Quit: Connection closed for inactivity]
vtunka has quit [Quit: Leaving]
viking has joined #jruby
viking has joined #jruby
mitchellhenke has joined #jruby
vtunka has joined #jruby
<headius>
good morning!
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] enebo pushed 7 new commits to master: http://git.io/1BeKFA
<JRubyGithub>
jruby/master 165e845 Thomas E. Enebo: First move towards specialized instrs during IRBuild time
<JRubyGithub>
jruby/master 34acdfe Thomas E. Enebo: Immutable CurrentScope can be shared. 500-700k reduction for Rails console (28k instances to 5).
<JRubyGithub>
jruby/master 78c0a23 Thomas E. Enebo: Immutable TemporaryCurrentModuleVariable can be shared. ~500k reduction for Rails console (24k instances to 5).
JRubyGithub has left #jruby [#jruby]
enebo has joined #jruby
<headius>
wooo
<headius>
enebo is pushing
<enebo>
happy fucking new year
<enebo>
so that is about 20+ more M off of memory usage
<headius>
awesome
<enebo>
including lazy IRBuilding of methods
<headius>
how far off are we now?
<enebo>
well I think 108M vs 57M
<enebo>
but we were 185M and I still have not deleted any datastructures
<enebo>
My pre goal is <100M
<enebo>
Once I get rid of specializeForInterpretation altogether I think I can swap things so JIT clones CFG instead of interp
<enebo>
That should get rid of about 6% of our memory usage
<headius>
ok
<enebo>
specializeForInterp on callinstr is done
<enebo>
attrassign and noresult are still pending
<enebo>
the whole factory of specialized versions of a single instr feels icky to me but it is no worse than before
<headius>
so, we should talk priorities for getting pre out
<headius>
that fix last night makes it possible for rails to generate apps correctly again
<imperator>
happy new year enebo :)
<enebo>
imperator: 2015 will be glorious
<imperator>
oh? why's that?
<enebo>
imperator: well 9k will be out for one :)
<imperator>
huzzah!
<enebo>
That will make me feel better about this year
<subbu>
headius, so, since that bug had been around for ever, how did rails ever generate apps correctly till recently (which i think it did).
<headius>
subbu: that's a good question, I'm not sure
<enebo>
I definitely have generated and started apps but I did not use the app much past running it
<enebo>
which version would not generate?
<headius>
the code that triggers this is not in rails 3.1, and may not be in earlier versions of thor
<enebo>
I mean I generated one from 4.16
<subbu>
so, maybe a gem update triggered this then
<enebo>
4.1.6 and I generated my app I am playing with last week
<headius>
that seems like the most likely answer right now, since this code was even the same in 1.7 branch
<headius>
enebo: app generated with master last week?
<enebo>
yes
noop has quit [Ping timeout: 256 seconds]
<enebo>
with 4.1.6
iamjarvo has joined #jruby
<enebo>
which perhaps means something after 4.1.6 somehow changed
<enebo>
Or something since I generated it last broke
<enebo>
but it seems like that bug is longstanding
<headius>
oh fwiw, this was not reported failing on 4.1 and passed on rails 3.1... just failure 3.2, 4.0, 4.2
<subbu>
i am getting build errors ..
<subbu>
[ERROR] /home/subbu/jruby/core/src/main/java/org/jruby/truffle/translator/TranslatorEnvironment.java:[16,45] error: cannot find symbol
<subbu>
[ERROR] symbol: class ReadLevelVariableNodeFactory
<subbu>
with mvn clean; mvn
<subbu>
something i should blow out from mvn cache? or update?
<headius>
hmm
<chrisseaton>
on master?
<headius>
wasn't failing last night so it would have to be very recent
<chrisseaton>
I'll just try pulling
<subbu>
yes, master on my laptop.
<enebo>
subbu: weird I did a pull —rebase before my push and did a clean build
<subbu>
i assume it is just something corrupted locally.
<chrisseaton>
if you have a build error with a clean build, maven sometimes seems to get confused and can't rebuild the generated sources next time around
<subbu>
ya .. something locally corrupted it appears.
skade has quit [Quit: Computer has gone to sleep.]
<subbu>
i dont havae generated sources.
<enebo>
I did not miss any new files so it is not that
<headius>
seems ok here too
<subbu>
any clue how i can get mvn to behave?
<enebo>
subbu: hit it with a newspaper
<enebo>
yeah just built for me
<enebo>
perhaps remove core/target?
<enebo>
I assume clean does that though
mitchellhenke has quit [Quit: Computer has gone to sleep.]
<subbu>
ok, will worry about this later .. i have a fix for #2409 .. but will push later .. got to get breakfast and get back to work work ..
<headius>
work work!
<subbu>
:)
tenderlove has joined #jruby
<headius>
my first priority is a few compiler specs for case/when forms
subbu is now known as subbu|breakfast
<enebo>
hah…I keep forgetting to -X-C my rails runs and I think what why did the memory go back up
<enebo>
but good news is ralis console with JIT is 111M
skade has joined #jruby
<enebo>
which is not great but considering -X-C is 108M it is pretty good
subbu|breakfast has quit [Remote host closed the connection]
skade has quit [Client Quit]
mitchellhenke has joined #jruby
subbu has joined #jruby
Hobogrammer_ has quit [Ping timeout: 244 seconds]
<headius>
yeah that's not too much worse :-)
<headius>
this is total process size, right?
calavera has joined #jruby
<headius>
or heap?
skade has joined #jruby
mcclurmc has quit [Remote host closed the connection]
<enebo>
used heap
<enebo>
on Java 8
<enebo>
so meatspace is in effect :)
<enebo>
headius: one weird thing. I just removed a boolean field from instr expecting to see a word less per instance and jvisualvm does not seem to be showing that
<enebo>
headius: do you know if Java 8 somehow is not sharing words for bools or small values?
<headius>
what's the class?
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] eregon pushed 2 new commits to master: http://git.io/m7Cv6w
<JRubyGithub>
jruby/master 52351c4 Benoit Daloze: Remove tainting from patterns in RubyDateFormatter....
<JRubyGithub>
jruby/master 41d41d1 Benoit Daloze: [Truffle] Implement Dir#delete so MSpec does not complain when running core specs.
JRubyGithub has left #jruby [#jruby]
<headius>
it's probably a memory alignment thing
<enebo>
Instr.java
<enebo>
I thought all fields got their own word for ones which were a word or less in size
<headius>
I believe it will attempt to pack some fields together
<headius>
what does it say the size is?
<enebo>
well exactly the same but I am looking at a single derived class LineNumberInstr
<headius>
yeah, I want to know the number :-)
<enebo>
well I am looking at total for all linenumberinstr and comparing it against pre-patch
<enebo>
So it is not real helpful to share it
<headius>
yeah I'm just trying to tie the reported per-object size with the layout
<enebo>
well I would need to do a heap dump for that I guess
<headius>
if the reported per-object size is in a multiple of 64 bits, then these booleans could be sneaking into the 32-bit object reference's chunk
<headius>
I don't know exactly how hotspot packs fields
<enebo>
headius: I am more curious about this because I thought they did not ever do this
pchalupa has quit [Quit: Leaving]
<enebo>
headius: but they clearly are now
donV has quit [Quit: donV]
<enebo>
headius: unless jvisualvm is not reporting size accurately on a heap sample
<headius>
that's possible
<enebo>
headius: The actual graph itself is screwy but the sample table seems reasonably consistent
baroquebobcat has quit [Quit: baroquebobcat]
<headius>
enebo: try -XX:-UseCompressedOops
<headius>
should be + by default
<enebo>
headius: I am googling
<enebo>
headius: I am guessing some platforms they do optimizations and some they do not
<enebo>
headius: My knowledge on this is out of date
<headius>
I do know compressed oops uses 32-bit references for objects rather than 64-bit
<headius>
mapped internally to appropriate 64-bit pointer
<enebo>
yeah no compressed makes the line number instrs about 500k larger
<headius>
with that flag you may see the impact of the boolean
<enebo>
yeah perhaps
<headius>
since the object reference before them will get full 64 bits
<enebo>
So long as I know some cases packs values I am happy with that
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius pushed 1 new commit to master: http://git.io/_jIQbQ
<JRubyGithub>
jruby/master e9e06aa Charles Oliver Nutter: MRI loads unicode_normalize.rb on boot now.
JRubyGithub has left #jruby [#jruby]
multibot_ has quit [Remote host closed the connection]
<nirvdrum>
enebo: I'm looking to roll some more commonality between the two String implementations into an interface. Are you at all worried about call performance through an interface?
multibot_ has joined #jruby
<headius>
nirvdrum: not generally
<enebo>
nirvdrum: I guess it depends on how coarse it is
<headius>
and how many implementations are live at once
<headius>
1-2 no problem...more than that starts to become vtable dispatch even after JIT
<enebo>
nirvdrum: if you call into an interface per byte then it may be a problem :)
<headius>
that too :-)
<enebo>
but that is probably less an issue with interfaces at that point :P
<headius>
nirvdrum: go for it
<headius>
we can sort out those details later if necessary
<enebo>
yeah we can always revert it!
<headius>
bam!
<nirvdrum>
Heh. No. I already have this CodeRangeable interface in there. But some of these static calls now take { CodeRangeable, ByteList, strLen }. It's effectively the same object 3 times.
<headius>
ok
<enebo>
nirvdrum: It will be cool to see how this develops
<nirvdrum>
I can clean up a fair bit by having a notion of something with a code range and a ByteList. And I can have the strLen calculated from those.
<enebo>
nirvdrum: I cannot say our String impl is so great but mostly because we took a more port it approach
<nirvdrum>
enebo: It's a mess right now, but it's coming along. Patterns of common usage are developing now that I've done a few.
<enebo>
nirvdrum: One thing to consider as well although I feel leery about this suggestion
<nirvdrum>
enebo: The trade-off is there's no longer a 1:1 mapping to MRI's C in some areas. But some of that stuff was from 1.8.6, so I'm not really sure there was anymore anyway.
<enebo>
nirvdrum: MRI CR values are somewhat driven from their memory layout
tcrawley is now known as tcrawley-away
<enebo>
nirvdrum: So we could for example add a length cache since we are not constrained in the same way
colinsurprenant has quit [Ping timeout: 250 seconds]
mcclurmc has joined #jruby
<enebo>
nirvdrum: I think CR is somewhat an opt based on what they could do with a single value
<nirvdrum>
That's stored in the object header right now. I didn't want to dick around with that :-)
<nirvdrum>
It's completely stubbed out in Truffle at the moment.
<enebo>
nirvdrum: My only reason for saying leery is that if you look at methods like cat19 you can see how intertwined CR is with behavior
<nirvdrum>
It's used for single byte optimizations and some code fails if the CR is reported as broken.
<enebo>
nirvdrum: ok
<enebo>
yeah
<enebo>
nirvdrum: It took ages to fix CR-related issues
<nirvdrum>
Rather than yank all that for Truffle, I figure I would just lie and always report a valid CR. But by having it baked into the interface, we get all that crazy validation for free once CR is implemented properly.
<headius>
it's a somewhat hacky fix but I don't have a better one for random-access
<headius>
without CR, all random access into even single-byte strings would be O(n)
<enebo>
headius: no I think CR for 7bit ascii is fine but for non-7bit caching length+ is an interesting idea
<headius>
agreed...especially since we have to traverse string to calculate CR :-)
<enebo>
yeah
<enebo>
that is my general point that MRI did not go beyond 7bit because of their memory layout
<headius>
could be
<enebo>
perhaps also to not explode code paths too…I guess I am speculating a little bit
<enebo>
I do know someone told me length was not cached because they had no room for it
<nirvdrum>
Well, hopefully this stuff will be easier afterwards. RubyString is getting slimmed down a lot. StringSupport is getting a bit ridiculous, but I can divide and conquer over there.
<enebo>
headius: It is prepareForInterpretation not getting called
<enebo>
headius: This is something I would like to solve anyways
<headius>
reverted it without commit so I won't accidentally push
erikhatcher has joined #jruby
<enebo>
headius: either way. I can also reapply but I had to change one of thing for this commit so I guess I should not be surprised this broke threshold 0
Usuario_ has quit [Read error: Connection reset by peer]
<headius>
our IR prep logic is still pretty fragile
erikhatcher has quit [Client Quit]
<enebo>
headius: yeah I want to make it less so since I want to change responsibility of JIT to clone CFG instead of interp
<enebo>
headius: it will halve the number of instrs we have in the system
<enebo>
ultimately I would like to not modify instrs at all but I think that is not achievable short term
baroquebobcat has joined #jruby
paulswilliamsesq has quit [Ping timeout: 245 seconds]
<headius>
ok
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] enebo pushed 2 new commits to master: http://git.io/GKGyIA
<JRubyGithub>
jruby/master 7623840 Thomas E. Enebo: Remove unused and somewhat uncommon hasUnusedResult marking
<JRubyGithub>
jruby/master 8a84610 Thomas E. Enebo: Make sure methods will prep for interp if they have not processed AST yet
JRubyGithub has left #jruby [#jruby]
<enebo>
headius: fixed spec:compiler
<headius>
cool, thanks
<finch>
rtyler: going to be jgitting it soo, last time I worked on it I made some really good progress
<headius>
I have some specs landing shortly
<finch>
However I haven't done a day of work since the 23rd so I haven't the faintest idea of what's going on
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius pushed 1 new commit to master: http://git.io/ADpY1Q
<JRubyGithub>
jruby/master 4e96624 Charles Oliver Nutter: Add compiler specs for case/when forms weakly covered elsewhere....
<headius>
come on, guys...jacoco's report generation only works as an ant task or maven plugin?
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] thomaswue opened pull request #2425: Further Truffle-related clean ups and simplifications. (master...master) http://git.io/bTUj3w
JRubyGithub has left #jruby [#jruby]
towski has quit [Ping timeout: 245 seconds]
<lopex>
nirvdrum: howdy
<nirvdrum>
lopex: Hey. I'm just about to eat lunch and I have a bit more reading to do. I'm trying to understand string length code.
<nirvdrum>
It seems RubyString and StringSupport have competing implementations.
<nirvdrum>
I'd like to simplify if I can.
mcclurmc has quit [Remote host closed the connection]
<lopex>
strLengthWithCodeRange ?
shellac has quit [Quit: Ex-Chat]
benlovell has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] tobiassvn opened issue #2426: NullPointerException when trying to assign UnboundMethod using define_method & to_proc http://git.io/Tu3Anw
<nirvdrum>
There's a form of strLengthWithCodeRange that has the same signature as a strLength. And the latter doesn't use the former.
imperator has quit [Quit: Valete!]
<nirvdrum>
But RubyString#strLength will use StringSupport.strLengthWithCodeRange
<lopex>
nirvdrum: appears to be uses by sublength
<lopex>
the other one
<lopex>
*to be using
<lopex>
nirvdrum: mostly likely cr related, since substring might end up with different cr
vtunka has quit [Quit: Leaving]
<headius>
well, I have a jacoco report
erikhatcher has joined #jruby
marr has joined #jruby
<headius>
and a Ruby script to generate it
<headius>
hmm...numbers don't seem to match reality yet...interpreter has zero coverage
<subbu>
headius, chrisseaton with enebo's tip, i figured out that mvn was barfing because of a compilation error in my patch. i had to checkout master without my patch, do a clean build and rebuild my patch to get mvn to tell me about the compilation error instead of throwing up.
<headius>
ahhh ok
mkristian has quit [Ping timeout: 244 seconds]
<subbu>
i had a new file lurking around in my repo earlier (git stash doesn't stash unadded files which i forgot about)
donV has joined #jruby
mcclurmc has joined #jruby
<nirvdrum>
lopex: How good are the tests for all this? I'm going to be careful, but there's a lot going on.
<lopex>
nirvdrum: well, the specs
<lopex>
nirvdrum: and blame mri
<lopex>
nirvdrum: any inconsistent behaviour will be covered by the tests as well :)
<lopex>
nirvdrum: like sometimes throwing exceptions on bad input and so on
skade has quit [Quit: Computer has gone to sleep.]
pietr0 has joined #jruby
benlovell has quit [Ping timeout: 250 seconds]
<lopex>
nirvdrum: also, being too smart about this will hinder future mri tracking (was bitten by that in the past)
<nirvdrum>
lopex: Great. I'd really like to get this down to one hierarchy of strLength calls if I can.
<lopex>
nirvdrum: but I'm not discouraging you :)
<nirvdrum>
lopex: Is that a big deal. A lot of this code hasn't been touched since 1.8.6. I'm not sure how closely we're tracking MRI with it.
<lopex>
nirvdrum: since 1.8.6 ?
mcclurmc has quit [Remote host closed the connection]
<lopex>
nirvdrum: 1.8.6 had non of that
<lopex>
*none
<headius>
ahhh there we go
<headius>
you have to generate report using the exact same class fles as you used at runtime
<nirvdrum>
lopex: Yeah, I guess I had the dates wrong. 1.9 is older than I realized.
<lopex>
nirvdrum: I've been doing some review two years back, so it might be reasonably up to date
<headius>
enebo: spec:compiler hits 60% of instructions in org.jruby.ir.interpreter.Interpreter
<headius>
I think these reports will be useful
<headius>
JVM instructions, that is
mkristian has joined #jruby
mcclurmc has joined #jruby
<headius>
yeah, these make sense...only thing in main interpreter loop not hit is the int/float specialization
<headius>
and some debugging/logging code
<nirvdrum>
lopex: Well, if you object to unifying let me know before I get too far. I'm really just trying to avoid the nested levels of calls. I'm trying to move stuff out of RubyString and into StringSupport. It's a bit hard when 8 methods need to be moved at once.
<lopex>
nirvdrum: just make sure cr tainting happens the same way
<lopex>
nirvdrum: afaik it was main pain since it doesnt show up in tests directly
<lopex>
nirvdrum: I know it should be simple, but mri isnt
<nirvdrum>
Will do. I'm not really stripping anything out. Just changing signatures a bit and maybe inlining if it makes sense.
<lopex>
nirvdrum: you can also consult rubinius sources
donV has quit [Read error: Connection reset by peer]
<lopex>
inlining ?
anaeem1_ has joined #jruby
anaeem1 has quit [Read error: Connection reset by peer]
erikhatcher has quit [Quit: erikhatcher]
<lopex>
nirvdrum: common paths (the paths without utf16 specializations for example) should ba as small as possible
<nirvdrum>
Replacing some of the simpler definitions with the counterparts they delegate to.
donV has joined #jruby
<headius>
lopex: I don't believe rbx has any code range in their string
<lopex>
headius: on purpose ?
<headius>
well, they started before encodings, and never added it...I can't speak to motivation
<enebo>
and almost all of those were weird Time/DateTime bugs
<headius>
what were we on 1.7? :-)
tlarevo has quit [Remote host closed the connection]
<enebo>
Good question but not relevent to trying it now since 1.7 won’t work now
<bbrowning>
travis acting weird w/ jruby? I'm seeing it use jruby 9k for my jruby-19mode matrix
<headius>
bbrowning: that's the travis bug asarih is working on
<enebo>
bbrowning: yeah asarih has an issue on it
<bbrowning>
ahh ok
tlarevo has joined #jruby
<headius>
there was a glitch in rvm where it started building master when we reached 1.7.18
<headius>
some matching issue with MRI 1.8
<enebo>
It is like why there is no Windows 9
<headius>
heh yeah :-)
<enebo>
We’re just like Windows
<enebo>
on rvm
<nirvdrum>
We didn't even need to get to 2.10 for me to enjoy myself.
<headius>
ok, I have MRI test suite numbers
<enebo>
yay
<headius>
coverage of core classes is better...61% of instrs in org.jruby
<headius>
that's not surprising I suppose
<skroon>
headius: well I wanted to migrate a big ruby app couple of years ago to Jruby, but I failed for a couple of reasons you mentioned in the talk I saw from you :)
<JRubyGithub>
[jruby] chrisseaton pushed 1 new commit to master: http://git.io/HCJaNw
<JRubyGithub>
jruby/master 6e85813 Chris Seaton: [Truffle] Remove old String#include? node.
JRubyGithub has left #jruby [#jruby]
<headius>
sigh, travis fail
yfeldblum has quit [Remote host closed the connection]
<electrical>
that's why i use jenkins :p
<chrisseaton>
Is Jenkins really better than Travis? I haven't had a chance to use it yet - but it looks like a 90s CMS
<nirvdrum>
Well, with Jenkins you control the operating environment.
<enebo>
chrisseaton: Jenkins is much more flexible in what you can do but Travis is a service and Jenkins is software so a better question may be is cloudbees bettre than travis
<enebo>
chrisseaton: At least I hope I did not misrepresent what I think you mean :)
<enebo>
chrisseaton: to me I prefer Travis in that I don’t have to try and manage it (we had our own Jenkin’s install for years). I won’t weigh in on the cloudbees vs Jenkins thing
<enebo>
but if anyone from travis is listening joshk (hmm not here) and asarih. Add scheduled jobs please :P
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] ahorek opened issue #2429: Can't run windows build (9000) http://git.io/bM8pwQ
JRubyGithub has left #jruby [#jruby]
<chrisseaton>
enebo: can't you just run scheduled jobs on every push? have they ever asked you to stop using so much build time?
rsim has joined #jruby
<enebo>
chrisseaton: well we want a nightly and we have to use cloudbees to accomplish that
<chrisseaton>
enebo: can't you just have a build per push?
<enebo>
chrisseaton: travis’s most requested feature is adding scheduled jobs so I am just ribbing them (or prodding them)
<chrisseaton>
no build building on a night nobody's pushed
<enebo>
chrisseaton: well we could just keep replacing and pushing out every build to S3 I guess?
<enebo>
chrisseaton: I think our goal though was to not just push the build but also do some other once a day testing due to time constraints
<headius>
I should expand coverage to include all the classes we support, like joni and jcodings
<enebo>
chrisseaton: The time constraints is difficult for us since we have to finish our test runs in <50 mins on travis
<headius>
chrisseaton: jenkins UI is terrible, agreed
<enebo>
chrisseaton: which is a different issue but it is our issues
<enebo>
headius: except the log scrolls right
<headius>
that's true
<headius>
and their log snippit shows the BOTTOM, which is what everyone's interested in
x1337807x has joined #jruby
<enebo>
headius: except josh :)
rsim1 has quit [Ping timeout: 265 seconds]
<enebo>
headius: although he has a good point. He likes seeing very top of log because he wants to see which job he is looking at
<headius>
sure
<enebo>
headius: which I can see that someone may want that but I would rather see the last lines where the error is
<enebo>
not sure why I am directing this at you
<enebo>
:)
<headius>
preaching to the choir, brother
<enebo>
I am passionate about the bottom logging, collapsable left project view, and making the fucking hamburger menu an actual menu (or replacing that graphic) :)
<enebo>
but I have told them many time (I think 2-3 too many times) my desires so I will just need to hold my tongue at this point
<asarih>
enebo: there is a secret sauce we are working on that makes this possible.
<asarih>
enebo: this = periodic builds
<asarih>
still a secret, though.
<enebo>
ahahah yeah asarih I was going to aks about “this” :)
<JRubyGithub>
jruby/master ba8a97d Thomas E. Enebo: No more specializeForInterpretation. All specialization is part of IRBuilding
JRubyGithub has left #jruby [#jruby]
paulswilliamsesq has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
tlarevo has joined #jruby
camlow3__ has quit [Remote host closed the connection]
<headius>
woo
benlovell has quit [Ping timeout: 245 seconds]
nanoyak has quit [Remote host closed the connection]
nanoyak has joined #jruby
erikhatcher has joined #jruby
mister_solo has joined #jruby
skade has joined #jruby
<lopex>
numbers?
benlovell has joined #jruby
nanoyak has quit [Remote host closed the connection]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/orUpzw
<JRubyGithub>
jruby/master 8ced6d2 Kevin Menard: Moved the private variants of RubyString#strLength to StringSupport as static methods.
JRubyGithub has left #jruby [#jruby]
nanoyak has joined #jruby
erikhatcher has quit [Quit: erikhatcher]
kares has quit [Ping timeout: 240 seconds]
yfeldblu_ has quit [Remote host closed the connection]
<skroon>
headius: do you know btw, about better_errors not working with 9000? I remember there was something with binding_of_caller
mitchellhenke has quit [Quit: Computer has gone to sleep.]
camlow325 has joined #jruby
camlow325 has quit [Remote host closed the connection]
mitchellhenke has joined #jruby
x1337807x has joined #jruby
camlow325 has joined #jruby
camlow325 has quit [Remote host closed the connection]
jomendoz has joined #jruby
<headius>
skroon: I had not heard that, no
benlovell has quit [Ping timeout: 245 seconds]
<headius>
it's not too surprising...I'm not sure the binding_of_caller code has been updated to the new 9k runtime
<jomendoz>
Hello! Does anybody has encountered the UnsatisfiedLinkErrorr while using activerecord-jdbcsqlite3-adapter? I haven't found any solutions to the problem in a just plain new Rails 4 project.
<jomendoz>
I developed a project on an ArchLinux machine, rvm, jruby 1.7.18 and activerecord-jdbcsqlite3-adapter 1.3.13 without problems. But I recently moved to Ubuntu Trusty 14.4 and it just prints this line 'Java::JavaLang::UnsatisfiedLinkError (org.sqlite.core.NativeDB._open(Ljava/lang/String;I)V): org.sqlite.core.NativeDB._open(Native Method)...'
<mpapis>
jomendoz, gist it all => gist.github.com
<headius>
jomendoz: interesting error...perhaps you don't have sqlite3 installed?
<headius>
it looks like it can't find a native library it needs
camlow325 has joined #jruby
<jomendoz>
Yes, actually since trusty is a long term supported ubuntu, I upgraded to version 3.8.7 (latest) and nothing changes.
<subbu>
headius, so for issue 2409 . i have interp working correclty and looks like only part jit ifxes are done .. would you be okay looking at the rest of the fixes for jit? i can also do it later tonight or tomorrow morning. you can say no :)
<headius>
this claims that there's an issue with the driver on 64-bit
x1337807x has quit [Ping timeout: 256 seconds]
x1337807x has joined #jruby
skroon has quit [Quit: leaving]
<headius>
jomendoz: can you make a simple script to reproduce it? You could try passing JVM arg -verbose:jni to see it loading those native libs
<headius>
JAVA_OPTS=-verbose:jni
<headius>
subbu: looking
subbu has quit [Ping timeout: 244 seconds]
<jomendoz>
headius: hmm yes, I've already seen that post but, I tried with version 3.8.2 (ubuntu's trusty) and outputs the same. Since my arch box has 64 bits and used the lastest of everything it seems to my quite odd the issue
<headius>
do you have a lib64 dir? I ran into some issues with another native library on RHEL with its /usr/lib64
<headius>
long story short, something is preventing the native side from loading, and there's a number of possible explanations
x1337807x has quit [Client Quit]
nanoyak has quit [Remote host closed the connection]
elux has quit [Quit: Bye!]
yarker has quit [Max SendQ exceeded]
nanoyak has joined #jruby
mitchellhenke has quit [Quit: Computer has gone to sleep.]
<headius>
this shouldn't be hard for us to reproduce given a matching OS
<jomendoz>
I've been using Rails and Rake commands to do work with JRuby. So, for example, when I tried to pull a git repo and create the db, rake db:migrate would spit the error for the first time.
<headius>
you might want to report this to the activerecord-jdbc project too, though...this isn't really a JRuby bug
<headius>
it isn't really an ar-jdbc bug either, but they're closer to it
<headius>
start an issue there and ping @headius
<jomendoz>
rails server could launch webrick normally but when requesting host/resources would crash with activerecord model usage
<jomendoz>
ok,
<headius>
better to capture these links and discussion in an issue
<jomendoz>
In activerecord-jdbc-adapter README, the feedback sections pointed me to this freenode, sorry for being misguided :P
subbu has joined #jruby
nanoyak has quit [Quit: Computer has gone to sleep.]
nanoyak has joined #jruby
elia has joined #jruby
Aethenelle has quit [Quit: Aethenelle]
yfeldblum has joined #jruby
<subbu>
headius, ok, will push then.
anaeem1_ has quit [Remote host closed the connection]
anaeem1 has joined #jruby
elia has quit [Client Quit]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] subbuss pushed 1 new commit to master: http://git.io/iaOmNw
<JRubyGithub>
jruby/master 29371d9 Subramanya Sastry: Fix #2409: Splat --> BuildSplatInstr + Splat (for use in call args)...
JRubyGithub has left #jruby [#jruby]
mitchellhenke has quit [Quit: Computer has gone to sleep.]
<subbu>
headius, ah . i think i know what the problem might be .. a stray context load that i forgot to remove.
viking has quit [Remote host closed the connection]
jomendoz has quit [Quit: Page closed]
<subbu>
saw that as soon as I saw the diff on github.
<JRubyGithub>
[jruby] nirvdrum pushed 5 new commits to master: http://git.io/FY8LIg
<JRubyGithub>
jruby/master 7bc9317 Kevin Menard: Cleaned up parameter lists in StringSupport by introducing a new common way to get a ByteList from a string....
<JRubyGithub>
jruby/master 65d7bd8 Kevin Menard: Added more methods to CodeRangeable interface to consolidate code.
<JRubyGithub>
jruby/master dbc1094 Kevin Menard: [Truffle] Implement String#encode with an Encoding argument....
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/V2P4aA
<JRubyGithub>
jruby/master 863a2fa Kevin Menard: [Truffle] Retagged a failing Time spec.
JRubyGithub has left #jruby [#jruby]
<headius>
calling it a night folks, ttfn
nanoyak has quit [Quit: Computer has gone to sleep.]
nanoyak has joined #jruby
rsim has quit [Quit: Leaving.]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/V1TL1Q
<JRubyGithub>
jruby/master ca55072 Kevin Menard: Simplify calls to StringSupport.index by calculating arguments rather than passing them.
JRubyGithub has left #jruby [#jruby]
nanoyak has quit [Quit: Computer has gone to sleep.]
nanoyak has joined #jruby
elia has quit [Quit: (IRC Client: textualapp.com)]