<GitHub186>
[jruby] eregon pushed 1 new commit to truffle-head: https://git.io/vX5nA
<GitHub186>
jruby/truffle-head 20cfd07 Benoit Daloze: Handle no supplementary groups in grpowned? specs
vtunka has joined #jruby
pawnbox has quit [Ping timeout: 258 seconds]
tcrawley-away is now known as tcrawley
vtunka has quit [Quit: Leaving]
pawnbox has joined #jruby
claudiuinberlin has quit [Remote host closed the connection]
claudiuinberlin has joined #jruby
vtunka has joined #jruby
claudiuinberlin has quit [Ping timeout: 256 seconds]
zacts has joined #jruby
vtunka has quit [Quit: Leaving]
pawnbox has quit [Remote host closed the connection]
<GitHub165>
[jruby] eregon pushed 2 new commits to truffle-head: https://git.io/vX5zI
<GitHub165>
jruby/truffle-head 1a6ee07 Benoit Daloze: [Truffle] No need to lower when specializations take a long.
<GitHub165>
jruby/truffle-head 8bc2092 Benoit Daloze: [Truffle] Make sure LangInfo.CODESET is defined.
yahonda has joined #jruby
claudiuinberlin has joined #jruby
pawnbox has joined #jruby
claudiuinberlin has quit [Ping timeout: 260 seconds]
Aethenelle has joined #jruby
vtunka has joined #jruby
vtunka has quit [Quit: Leaving]
enebo has joined #jruby
<GitHub42>
[jruby] axfcampos opened issue #4319: JRuby can not interpret keyword argument when placed after positional argument in block https://git.io/vX5KM
<headius>
eregon: seems unlikely anyone would have accessed those directly, right?
maclover7 has joined #jruby
<headius>
I kinda wish those weren't public, but I could see someone using this across platforms
maclover7 has quit [Client Quit]
<headius>
meh, whatever
Aethenelle has quit [Quit: Aethenelle]
<GitHub105>
[jruby] headius created thread_report_and_string_charsequence (+2 new commits): https://git.io/vX5Fb
<GitHub105>
jruby/thread_report_and_string_charsequence d00c89f Charles Oliver Nutter: Implement Thread{.,#}report_on_exception[=]....
<GitHub105>
jruby/thread_report_and_string_charsequence 11272b1 Charles Oliver Nutter: Implement CharSequence in RubyString....
Aethenelle has joined #jruby
<lopex>
wow this one is madness
<lopex>
headius: are the byte[] symbols going into 9.2 ?
bbrowning is now known as bbrowning_away
donV has quit [Ping timeout: 256 seconds]
kith has quit [Quit: kith]
<headius>
lopex: you say that as though we have it ready to go
<lopex>
headius: enebo said it should
<lopex>
I witnessed the conversation :P
<lopex>
headius: but I still dont have any idea how to do that
<enebo>
hahaha
<headius>
well...anything's possible :-)
<lopex>
ditch j.l.String just to reuse it once jdk9 is there
<enebo>
yeah I thought about trying to implement that for 9.2
<enebo>
If we don’t it will just keep dragging on
<enebo>
I am a little worried though :)
<lopex>
well, jdk9 String wont let you share anything though right ?
<headius>
it's as good a time as any
<lopex>
just a flag
<headius>
lopex: well you can't see the bytes anyway
<lopex>
but you want to from jruby perspective
<headius>
the main advantage to us *might* be that 7bit Ruby strings can be copied rather than transcoded
<lopex>
headius: btw, they promise non breaking changes right ?
<lopex>
yeah
<lopex>
for the most common case
<headius>
yes
<lopex>
sure
<lopex>
that a big win
<lopex>
headius: er, I responded to the copy thing
<lopex>
headius: but we can do click trick and do some cheating there ?
<headius>
heh, I suppose there's tricks that could be pulled :-)
<lopex>
haha
<lopex>
headius: but the Cliff's stuff is still there right ?
<lopex>
wrt maps etc
<headius>
yeah we are using one of them now
<nirvdrum>
headius: It looks like you caught up with eregon. But it seems the SunOS vs Solaris thing was only partially fixed.
<lopex>
it always boils down to storing "strings" in the source or the bytecode ins some what managable way
<lopex>
nirvdrum: almost 1k loc less
<headius>
nirvdrum: because the old SunOS constants were still there, eh?
<lopex>
does jruby in any way support hot downloading anything ?
<lopex>
like the jruby jigsaw
<headius>
nothing like that currently, no
<lopex>
headius: we still package transcoding tables no one ever used I bet
<nirvdrum>
headius: The Rakefile generator was using the FFI OS value, which is normalized. While the enum loader wasn't normalizing the OS name in the same way.
<nirvdrum>
So, while we generated new constant values, they weren't being picked up by the auto-enum loader.
tcrawley is now known as tcrawley-away
<lopex>
auto-enum loader ?
<headius>
nirvdrum: ahh bummer
<GitHub87>
[jruby] headius pushed 2 new commits to ruby-2.4: https://git.io/vX5jY
<GitHub87>
jruby/ruby-2.4 b1df1ba Charles Oliver Nutter: Remove report-on-GC to align with MRI 2.4. #4293...
<GitHub87>
jruby/ruby-2.4 e408638 Charles Oliver Nutter: Implement Thread{.,#}report_on_exception[=]....
shellac has quit [Quit: Leaving]
<GitHub11>
[jruby] headius pushed 1 new commit to ruby-2.4: https://git.io/vXdfu
<GitHub11>
jruby/ruby-2.4 f802cd0 Charles Oliver Nutter: Update compat version in mspec config.
<headius>
lopex: yeah that would be nice to fix
<lopex>
headius: wrt strings ?
<lopex>
headius: not sure which are you responding to
<lopex>
headius: as, the symbols, nvm
<lopex>
*ah
<headius>
the transcodign tables
<lopex>
those are ok afaik ?
<lopex>
headius: aah, the packaging
<lopex>
headius: any idea wrt lazy/ondemand downloading ?
thedarkone2 has joined #jruby
<lopex>
headius: wouldnt maven package do ?
<lopex>
but it would require careful splitting for the most common usecases
<headius>
hmm
<headius>
we could provide a tool to install them
<headius>
gem install jruby-extended-encodings
<headius>
problem is I don't know what we can remove
<headius>
we have users all around the world
<lopex>
the tool would require some serious security prerequisites
<lopex>
since once you';ve donwloaded jruby you have to have the same level of trust
<lopex>
or changing the license ?
<lopex>
since the license would have to include the "third party thingy"
<lopex>
I dont know
<lopex>
headius: am I bollocks ?
<headius>
yeah I don't know
<lopex>
headius: I thing your gem idea is better
<headius>
it wouldn't rewrite the jruby jar
<lopex>
*think
<headius>
just add jars to some place we load
<headius>
maybe you have to require something too?
<lopex>
headius: can gem download anything without use interaction ?
<lopex>
*user
<headius>
sure
<headius>
jar-dependencies already does that...you specify maven coordinates
<lopex>
I'd go for it then
<lopex>
yeah
<lopex>
problem, solved for me
<lopex>
so gem with a platform right ?
<headius>
yeah, but now the question is what we can remove
<headius>
someone will complain about almost anything :-)
<lopex>
Transcoder_Big5_WordArray.bin
claudiuinberlin has quit []
<lopex>
headius: I'd sort that for size
<lopex>
headius: those are hige
<lopex>
headius: too bad we dont have any download info
<lopex>
then we might have a guess
<headius>
I'm sure there's a way to generate some report from S3
<lopex>
Transcoder_Gbk_WordArray.bin ?
<lopex>
Korean ?
<lopex>
all those are huge
<lopex>
big5
<lopex>
hey
<lopex>
headius: lots of easy ones actaully
<lopex>
GB* my favourite
<lopex>
headius: is utf8mac really mac ?
<headius>
that's a weird one
<lopex>
headius: we can easily win 2.5MB
<lopex>
those are compressible though too
<lopex>
headius: those compress to 1.1MB for me locally alone
<lopex>
top 30 by size
<lopex>
from tables
<lopex>
headius: well the question is is it a win ?
<headius>
we used to have folks concerned about size but not much lately
<headius>
it could strip down for ruboto but we have not supported that in 9k yet
<lopex>
headius: the other thing is maybe there's some tables left in bytecode
<lopex>
which is enormous in size
mikhailvs has joined #jruby
<mikhailvs>
how do i convert a ruby array into a float[]?
<lopex>
in the bytecode from what I can tell is the char types, and simple folds
<lopex>
to_a(:float) /me guessing
<lopex>
or
<lopex>
er
<mikhailvs>
to_java(:float)?
<lopex>
yep, more like it
<mikhailvs>
that worked
<mikhailvs>
thanks man
<lopex>
or
<lopex>
to_java(:float[])
<lopex>
?
<lopex>
ok
<mikhailvs>
to_java(:float) worked
<mikhailvs>
i dont think :float[] would be a valid symbol
<mikhailvs>
:'float[]' would tho
<lopex>
headius: the bytecode size wrt arrays is a joke
<lopex>
is dup, push index, push value, array store
<lopex>
amazing
<nirvdrum>
lopex: Sorry, stepped away to eat lunch. The jnr-constant Enums lazily load platform-specific values.
<lopex>
even java literals compile to this
<lopex>
nirvdrum: hiya
<lopex>
nirvdrum: oh, the enum loader
<lopex>
nirvdrum: hey, how does truffle run sinatra/sequel thingy ?
<nirvdrum>
lopex: What are you referring to?
<lopex>
nirvdrum: runnablilty mostly
mikhailvs has left #jruby [#jruby]
<lopex>
nirvdrum: like very lightweight deps system
<nirvdrum>
lopex: Sorry, I mean what sinatra/sequel think are you asking about?
<lopex>
nirvdrum: these are the deps I'm running with
<lopex>
nirvdrum: very modest
<nirvdrum>
Oh, so your own app.
<lopex>
yeah
<nirvdrum>
We run Sinatra fine. But you're not going to get anywhere with Sequel I'm afraid.
<nirvdrum>
We can run the redis-rb client and postgres-pr.
<lopex>
ok
<jeremyevans>
nirvdrum: Sequel can use postgres-pr
<lopex>
hmm, I can help with running sequel
<lopex>
nirvdrum: what do you lack ?
<nirvdrum>
jeremyevans: Right. Sorry, I may have misspoken there. I haven't tried that yet.
<lopex>
nirvdrum: java integration ?
<nirvdrum>
lopex: Generally, C extensions and JDBC support.
<lopex>
nirvdrum: I only care for jdbc
<nirvdrum>
Pure Ruby drivers should work fine.
<nirvdrum>
jeremyevans: I did test with roda though and we run that fine.
<lopex>
nirvdrum: is truffle/ruby integration settled enough ?
<lopex>
like in jruby ?
<nirvdrum>
I'm not sure what you mean.
<lopex>
nirvdrum: java integration
<nirvdrum>
We don't have any Java interop at the moment.
<lopex>
nirvdrum: sequel relies on that for jdbc etc
<lopex>
bummer
<lopex>
ok
<lopex>
nirvdrum: is there any paths settled on that ?
<lopex>
like "it should be similar to jruby"
<nirvdrum>
Nothing at the moment. We've been focusing more on running Ruby.
<lopex>
sure
<lopex>
so too early
<nirvdrum>
It's possible we might support it via a theoretical Truffle Java implementation.
<nirvdrum>
But otherwise, we'd need do something reflection-based.
<nirvdrum>
It'd be nice to have, for sure. It's a major selling point of JRuby.
<nirvdrum>
I routinely use it like a Java shell.
<lopex>
nirvdrum: well, it may happen I'll get paid from having it being usable
<chrisseaton>
Our Java interpreter is in the works
<lopex>
but hey, hows the AR support in the truffle ?
<lopex>
doesn it need the same support ?
<lopex>
chrisseaton: ^^
<nirvdrum>
lopex: Petr has AR working with postgres-pr.
<nirvdrum>
But AR is basically a set of one-off DB adapters, so it's a difficult target.
<lopex>
nirvdrum: but it does not rely on JI ?
<nirvdrum>
As the AR-JDBC guys can attest to.
<lopex>
ok
<lopex>
I get it now
<nirvdrum>
lopex: No. It uses the pure Ruby postgres driver.
bbrowning_away is now known as bbrowning
<lopex>
yes
<nirvdrum>
I suspect it's not as fast as something that binds to libpq or uses JDBC. But it's functional.
<lopex>
I just hoped that if one uses any jdbc thingy it will rely on JI
<lopex>
ok
<nirvdrum>
lopex: Well, in theory we could write Truffle nodes for all the endpoints that AR or Sequel use. And those nodes are written in Java and can call freely into Java.
<nirvdrum>
But at 3 people, we really don't want to be maintaining a dozen different DB adapters.
<chrisseaton>
My current thinking is we want to support the same stack people use on MRI
<lopex>
wel, you just need a good JI for that right ?
<eregon>
> <headius> eregon: seems unlikely anyone would have accessed those directly, right?
<eregon>
they could before, if System.getProperty("os.name") reporeted SunOS, but they were outdated and uplicated with constants in solaris/
<eregon>
duplicated*
<chrisseaton>
Java interop and ext support is very important but we have finite people and time
<lopex>
nirvdrum: why specialize anything to AR or sequel ?
<lopex>
chrisseaton: hey, but it's the time weeee need pushing
<lopex>
:P
<lopex>
truffle needs a good JI right ?
<chrisseaton>
Yes and it'll have it
<nirvdrum>
lopex: It is a bit more than that though. Since JRuby is the only thing with JDBC support currently, both Sequel and AR-JDBC assume they're running in JRuby. And we don't support JRuby's extension API either.
<lopex>
nirvdrum: right
<nirvdrum>
It's not insurmountable. But it is a bit messy.
<lopex>
lots of api assumptions too
<lopex>
hm
<nirvdrum>
We're certainly not short of work :-)
<lopex>
whats the diff from graal view and hotspots current wrt that ?
<nirvdrum>
With respect to what?
<lopex>
bindings ?
<nirvdrum>
We don't have them, so I can't say :-)
<nirvdrum>
I suspect we'd cache the method lookup and be able to inline that. But I suspect we can't really PE the JDBC methods.
<lopex>
nirvdrum: oh I dindt mean bindings per se, just the ability to call a function
<eregon>
nirvdrum: we could PE through Java with TruffleJava :)
nicoulaj has joined #jruby
pawnbox has quit [Ping timeout: 252 seconds]
claudiuinberlin has joined #jruby
pawnbox has joined #jruby
rtyler has joined #jruby
pawnbox has quit [Ping timeout: 260 seconds]
tcrawley-away is now known as tcrawley
pawnbox has joined #jruby
kith has joined #jruby
mtoy has joined #jruby
<mtoy>
headiuis: 9.1.6.0 seems to have fixed my
<mtoy>
headius: "slow load from nested jars" problem ... but the bug is still open ... am i crazy or is it fixed?
<enebo>
lopex: we have had a wallstrreet recovery but we have lots of angry people who are being easily swayed
<lopex>
I'm very liberal leftist
<lopex>
but maybe I shuld change ?
<enebo>
lopex: and not get captured in bot logs?
<lopex>
haha
<enebo>
lopex: the world has not been happy but picking an outsider who is not really in populist interests will kick the drama down the road longer. In US I think Sanders had a chance at lighting an important fire but he will be too old in 4 years
<enebo>
lopex: in rest of world I am less certain (I AM MURICAN)
<lopex>
enebo: yeah what do you think about sanders like a year ago ?
<chrisseaton>
enebo: I didn't see you after your RubyConf talk but it was really good this year
<chrisseaton>
good, fresh look at JRuby
<enebo>
chrisseaton: thanks. I feel like we have a lot of value adds but we concentrate too much on being a better MRI (except startup time)
<enebo>
chrisseaton: MRI + JRuby I think is better together than apart
<enebo>
lopex: I think Sanders is nearly perfect in expressing political discontent for most american's
<enebo>
lopex: with that said I am a bit older and was raised in the notion that things like college should be free
<enebo>
lopex: they were when I was a child (at least for many US states for public ediucation)
<lopex>
enebo: go on
<enebo>
heh amaericans
<enebo>
typos excluded
<lopex>
dunno
<enebo>
lopex: the notion of no single-payer system in US is crazy
<enebo>
lopex: I feel like we have been conviinced that we are incapable of change and that everything that is happening is inevitable
<lopex>
well, if I dint know you fo example I'd have different (percepction of americans")
<enebo>
lopex: but as a nation we definitely have options. We are a pretty influential conountry after all. If we can’t who can?
<enebo>
lopex: so my take is that we have options but we have a strong narrative that we have no options.
<enebo>
lopex: All boats must sink and not rise
<lopex>
enebo: yes
<enebo>
lopex: I think this is false but it heavilty depends on social will
<enebo>
lopex: the battle of social will is our real problem in the US
<lopex>
what's that ?
<lopex>
"the will"
<enebo>
lopex: in world war 2 we had the concept of sacrifice for the betterment of the nation
<lopex>
we were ethe heroest then
<enebo>
lopex: it would be wrong to say that is a form of nationalism but we had a shared vision of where we wanted to go and we did it
<lopex>
but hey "sacrifice" ?
<lopex>
er, why
<enebo>
lopex: but two generations ago are frequiently the heroist
<lopex>
sure
<enebo>
lopex: yeah good question
<enebo>
lopex: sacrifice may or may not be needed
<lopex>
I dont doubt the heroes
<enebo>
lopex: but the fact that we felt the need to come together
<enebo>
lopex: I feel the US problem is lack of coming together when it makes sense and always opposing because it fits our tribe
<enebo>
lopex: interestingly I think dems and repubs are having a identity crisis
<lopex>
chrisseaton: did you learn about division 303 in high for example
<lopex>
enebo: the tribe is a bad word,
<lopex>
enebo: lol "leftist"
<enebo>
we have been told for years repubs will lose to demographics changes but dems is losing due to progressive/DLC-third way split
<enebo>
lopex: tribalism is alive and well in US
<lopex>
is it ?
<enebo>
but the two parties are rupturing
<enebo>
we might have 4 or more tribes in a 2 party system
<enebo>
I apologize to any JRuby people wanting tech talk but me and lopex can get crazy
<lopex>
haha
<lopex>
enebo: I agree with you
<lopex>
enebo: remember what I was on ther verge on climate denialist ?
<lopex>
enebo: "when'
<enebo>
lopex: yeah
<lopex>
enebo: I was quickly turned around by you
<enebo>
lopex: thanks for saying that. You see the artic warmth this year?
<enebo>
lopex: it has been really warm in MN this fall
<enebo>
lopex: but jet stream pushed north
<lopex>
enebo: I'm so ashamed event if it lasted a month
<lopex>
enebo: but hey, I needed a week to be turned around
<lopex>
and still ashamed
<enebo>
lopex: don’t worry we just need to convince more people
<enebo>
lopex: and in actuality we just need less people in the world…our biome is a bit overpopulated
<lopex>
no, no I'm trying to find a reason why that happened
<lopex>
enebo: but theres so manny issues
<lopex>
and one mind
<enebo>
lopex: yeah and many voices
<lopex>
enebo: do you study USGS in any ways ?
<lopex>
there's more and more satelites on the way
<enebo>
lopex: I would say that I do not directly no. I guess I have seen other things which correlate that multiple discpline see problems in different ways
<enebo>
historical changes between artic ice cores and oceanic cores give me confidence in historical changes. Anything modern showing strong devidations makes me trust they are not making stuff up
<enebo>
I do think one problem is modelling is difficult. Weather is hard enough in days much less years. I think that is what make denialism so powerful
<lopex>
enebo: doom
<enebo>
lopex: I think we both love science and would love to finance it. Anything which curbs financing research is suspect to me
<lopex>
curbs ?
<enebo>
lopex: tries to prevent
<lopex>
ah
<enebo>
lopex: anyways I should get off my soapbox before I am banned from the internet
<chrisseaton>
I went to a college rather than a high school, but I did learn about Squadron 303, and actually taught my troop about it because we were by where some of them were buried
<chrisseaton>
Presuming division 303 is the same as squadron 303
<lopex>
enebo: but it's more complcated bacause of epigenetics
<lopex>
chrisseaton: yeah
<enebo>
lopex: but can parentage affect epigenetics?
<lopex>
enebo: yes
<enebo>
lopex: I feel like each calcium can but I guess I don’t follow the rest
<enebo>
lopex: epigenetics ia fascinating though
<lopex>
chrisseaton: hey, in poland it's whole history
<lopex>
enebo: but not a myth anymore
<chrisseaton>
I used to work at Pirbright Barracks, and I think many of them are buried at Pirbight Cemetary
<lopex>
chrisseaton: in poland we were thought we almost saved englang
<lopex>
chrisseaton: all politics
<enebo>
lopex: chrisseaton: when I worked in Canada for 1/2 a year they told me about how US history os WW2 was wrong and Canadian involvement and how England conducted the war
<enebo>
It was massively different than how US taught history of WW2
<enebo>
made watching the movie Patton much more interesting
<lopex>
chrisseaton: just translate polish wiki of 303
<lopex>
chrisseaton: and you have no way to check the truth