robbyoconnor has quit [Quit: Konversation terminated!]
nateberkopec has quit [Quit: Leaving...]
josh-k has joined #jruby
<GitHub136>
[jruby] enebo pushed 5 new commits to inline: http://git.io/vEWO4
<GitHub136>
jruby/inline 3619ab1 Thomas E. Enebo: New Site interface for getting (and for now setting) siteId. All calls...
<GitHub136>
jruby/inline 7f255e9 Thomas E. Enebo: Inline with JIT need only linearizeBB after passes
<GitHub136>
jruby/inline f660230 Thomas E. Enebo: Start to clean up
mjc_ has quit [Ping timeout: 240 seconds]
mjc_ has joined #jruby
<GitHub27>
[jruby] subbuss commented on commit 61f8da4: Can this happen? ClassBody is the only non-singleton class body => for all other scope types that is being inlined into the host scope, ScopeModule will always have scopeModuleDepth > 1. So, this scenario should not happen and you should assert. http://git.io/vEWZx
<GitHub97>
[jruby] subbuss commented on commit 61f8da4: That assumption holds because you are never inlining a class body into something else. http://git.io/vEWn4
<GitHub90>
[jruby] subbuss commented on commit 61f8da4: There are 2 other places in IRBuilder where ScopeModule comes from. But, both of them reference [IRScope.getNearestModuleReferencingScopeDepth](https://github.com/jruby/jruby/blob/master/core/src/main/java/org/jruby/ir/IRScope.java#L315-L334) and for all non-module-body scopes, you have depth > 0. So, once again since you are inlining methods and blocks, depth cannot be zero in the scopes that are being inlined.
<chrisseaton>
This is our uber build that includes R, JS, JRuby etc, but you can use it as a normal JVM with Graal - use jre/bin/javao rather than bin/java though (explained in that wiki page)
<chrisseaton>
I'm seeing 2.5x Mandelbrot when using indy, which is almost too good to be true
<headius>
why oh why does your login not post when I hit enter, Oracle?
<chrisseaton>
So I'm running all the benchmarks on some stable hardware
<headius>
I almost always expect it to and then walk away from the tab while it does nothing
<chrisseaton>
travis 'still failing' messages should be directed at the person who originally broke the build
pitr-ch has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<headius>
indeed
<chrisseaton>
headius: looks like some are faster, some are actually slower - I'll have to open bugs for those
<headius>
yeah I'm getting a mixed bag
<headius>
most small benchmarks do seem faster though
pitr-ch has joined #jruby
thedarkone2 has joined #jruby
<chrisseaton>
I'll run the full set - with and without indy and trying the interpreter as well
<headius>
nice, thanks
<headius>
what's with the javao thing? "graal bug"?
<chrisseaton>
bin/java is a wrapper that loads all the other languages so you can do interop, but it uses JAVACMD, which trashes our JAVACMD and it all goes wrong from there
<headius>
ahhh ok
<chrisseaton>
javao is just a regular JVM with Graal enabled, no wrapper
pitr-ch has quit [Ping timeout: 256 seconds]
pawnbox has quit [Remote host closed the connection]
pawnbox has joined #jruby
<headius>
modified my local copy to just use javao
<headius>
subbu: I suspect block protocol may have messed up enebo's define_method optimization
<headius>
a non-capturing define_method is now 2x slower than a capturing define_method, when it should be almost as fast as a normal method
pawnbox has quit [Remote host closed the connection]
pawnbox has joined #jruby
skade has quit [Quit: Computer has gone to sleep.]
<GitHub63>
[jruby] headius pushed 2 new commits to master: http://git.io/vE8Ny
<GitHub63>
jruby/master 3e3be35 Charles Oliver Nutter: Revert "Revert "Symbol#to_proc procs' #parameters always return [[:rest]].""...
<GitHub63>
jruby/master 4f866e2 Charles Oliver Nutter: If it should error, make it error always.
<lopex>
numbers?
<lopex>
headius: I see no changes in encoding and trandscoding tables (apart form new transcoder being added)
<lopex>
wrt 2.3 master
<headius>
lopex: strange...NEWS says they updated unicode
<headius>
maybe the update didn't change anything
<lopex>
yeah
<lopex>
headius: that new trascoder doesnt require new impl
<headius>
what does it do?
<lopex>
ebcdic to iso-8859 and back
<headius>
Martin Dürst was interested in our port of the transcoding subsystem btw
<headius>
I believe he was the madman behind the scenes of m17n
<lopex>
yes
<lopex>
we tried to contact him back then
<headius>
lopex: yeah I think we'll wait until someone reports that we don't handle EBCDIC Right
<lopex>
headius: also, there's two new windows encodings
<headius>
that would be worth adding
gaustin_ has joined #jruby
<lopex>
headius: it's all on jcodings master
<headius>
nice!
<headius>
I can spin a release whenever you think it's ready
<lopex>
headius: also, simpified loading and got rid of that encoding list table
<lopex>
same for transcoders
<headius>
ahh excellent
<headius>
that was confusing
<lopex>
headius: I only ran -P test with that though
<headius>
we can run MRI suite with it in JRuby
<lopex>
headius: and that newline.c transcoder is back where it was
gaustin has quit [Ping timeout: 260 seconds]
gaustin_ is now known as gaustin
<GitHub187>
[jruby] headius pushed 1 new commit to ruby-2.3: http://git.io/vE8xu
<GitHub187>
jruby/ruby-2.3 0c49c86 Charles Oliver Nutter: Merge branch 'master' into ruby-2.3...
<headius>
"git checkout -" is my new favorite git feature
<lopex>
what does that do ?
<headius>
like cd -
<headius>
checks out previous branch
<lopex>
ah, should have guessed that
pitr-ch has joined #jruby
<GitHub191>
[jruby] nirvdrum pushed 1 new commit to truffle-head: http://git.io/vE4UR
<GitHub191>
jruby/truffle-head ab235df Kevin Menard: Merge pull request #3561 from jruby/rbx-2.7-upgrade...
<GitHub195>
[jruby] nirvdrum closed pull request #3561: [Truffle] Upgraded from Rubinius 2.5.6 to 2.7. (truffle-head...rbx-2.7-upgrade) http://git.io/vEWKX
<GitHub196>
[jruby] nirvdrum deleted rbx-2.7-upgrade at 8eb466e: http://git.io/vE4U7
<subbu>
those 4 of the ones i emailed you about y'day yes.
<headius>
yeah
<subbu>
which ones are odd .. you mean the regexp ones?
skade has quit [Ping timeout: 255 seconds]
<headius>
well they're all a bit surprising
<headius>
I haven't started looking into it
<subbu>
k
<subbu>
how do you print those oids in ruby? what is the method .. i've forgotten my ruby :)
<subbu>
[exec] Expected /OK/ (oid=11590) to be the same as /OK/ (oid=11592).
<headius>
object_id
<headius>
I expect that one is not reusing the same regexp
<subbu>
perhaps.
<headius>
I have not looked...I planned to get 2.3 branch greener today
<subbu>
i am in wmf work mode right not .. but if there are some spare cycles later in the evening .. will poke at the two regexp ones.
<subbu>
unless you get to them first after ruby 2.3
<headius>
sure no problem
<headius>
you doing anything for xmas?
<subbu>
nothing much .. neither of us are christian ..
<headius>
nice quiet time of year then
<subbu>
but, might go to some friend's house for a bit during the day. and otherwise read books and do nothing .. if such a thing is possible :)
<subbu>
yes, usually quiet time of the year for us.
<subbu>
i suppose your families get together for the occasion?
skade has joined #jruby
subbu is now known as subbu|breakfast
brauliobo has quit [Ping timeout: 246 seconds]
<headius>
xmas eve at Sara's dad's, xmas day at her mom's
<headius>
my family doesn't really celebrate
<headius>
ok, after five minutes staring at the parser, I can confidently say I still do not know how to fix anything in it
<headius>
I guess this spec gets tagged for enebo
skade has quit [Ping timeout: 245 seconds]
<GitHub82>
[jruby] headius pushed 1 new commit to ruby-2.3: http://git.io/vE4yx
<GitHub82>
jruby/ruby-2.3 0304cc7 Charles Oliver Nutter: Tag parser failure from 2.3 changes. #3479
<GitHub181>
[jruby] chrisseaton fast-forwarded truffle-0.10 from c390168 to c81c55e: http://git.io/vE4SN
<GitHub95>
[jruby] chrisseaton pushed 1 new commit to truffle-head: http://git.io/vE4Sp
<GitHub95>
jruby/truffle-head bbe61ef Chris Seaton: Merge branch 'truffle-0.10' into truffle-head
subbu|breakfast is now known as subbu
<headius>
chrisseaton: whenever you talk to graal folks about those slowdowns, copy me
<headius>
I'd love to see it be faster across the board and I know there's things it can do for us hotspot can't
blaxter has quit [Quit: foo]
ianderse has quit [Quit: Connection closed for inactivity]
<headius>
wow, what the heck
<headius>
trying to run mspec from the gem and I get a bunch of really wacky errors
<headius>
NoMethodError: undefined method `double' for #<Object:0x4164ab12>
<GitHub152>
[jruby] chrisseaton pushed 1 new commit to master: http://git.io/vE453
<GitHub152>
jruby/master b276f01 Chris Seaton: Merge branch 'truffle-0.10'
<chrisseaton>
headius: i'll keep you informed
<headius>
thanks...I have good contacts on the hotspot team but not on graal side
<chrisseaton>
the team is really just Doug Simon, Gilles Duboscq, Stefan Anzinger, Tom Rodriguez and Roland Schatz - you may already know some of them?