<eonwe>
This is inspired by the problems we have with Immutables-based builders. Having the implementations there package-private causes trouble if JRuby-jar is not in boot classpath (which seems kinda odd to me)
vtunka has quit [Quit: Leaving]
donValentin has joined #jruby
donV has quit [Ping timeout: 256 seconds]
vtunka has joined #jruby
<GitHub103>
[jruby] jmiettinen opened issue #4266: JRuby unable to find methods unless jar is in boot classpath https://git.io/vXRv2
<eonwe>
I created an issue out of it. I might've just misunderstood what is and is not supported.
pawnbox has quit [Remote host closed the connection]
claudiuinberlin has joined #jruby
temporalfox has joined #jruby
vtunka has quit [Quit: Leaving]
<GitHub49>
[jruby] eregon pushed 2 new commits to truffle-head: https://git.io/vXRkE
<GitHub34>
jruby/master c7e1459 Thomas E. Enebo: Fixes #4227. Missing unit :second for Process.clock_gettime
<GitHub27>
[jruby] enebo closed issue #4227: Missing unit :second for Process.clock_gettime https://git.io/vPr60
donV has joined #jruby
donValentin has quit [Read error: No route to host]
vtunka has joined #jruby
<GitHub195>
[jruby] enebo closed issue #4167: Create a different hierarchy of DynamicScope for use in startup interp + remove null checks in the existing hierarchy since they are redundant https://git.io/vi7v5
pawnbox has joined #jruby
deathy has quit [Ping timeout: 250 seconds]
asarih_ has joined #jruby
beawesomeinstead has quit [Ping timeout: 250 seconds]
andrewvc has quit [Ping timeout: 250 seconds]
mjc_ has quit [Ping timeout: 250 seconds]
Scorchin has quit [Ping timeout: 250 seconds]
amdprophet has quit [Ping timeout: 250 seconds]
phlebas has quit [Ping timeout: 250 seconds]
aemadrid has quit [Ping timeout: 250 seconds]
fidothe has quit [Ping timeout: 250 seconds]
mccraig_ has quit [Ping timeout: 250 seconds]
asarih has quit [Ping timeout: 250 seconds]
bruceadams has quit [Ping timeout: 250 seconds]
asarih_ is now known as asarih
zacts has quit [Ping timeout: 250 seconds]
Scorchin has joined #jruby
bruceadams has joined #jruby
amdprophet has joined #jruby
bbrowning is now known as bbrowning_away
deathy has joined #jruby
andrewvc has joined #jruby
fidothe has joined #jruby
beawesomeinstead has joined #jruby
mccraig_ has joined #jruby
pawnbox has quit [Ping timeout: 250 seconds]
aemadrid has joined #jruby
phlebas has joined #jruby
mjc_ has joined #jruby
pawnbox has joined #jruby
cprice has joined #jruby
andrewvc has quit [Ping timeout: 260 seconds]
Specialist has quit [Remote host closed the connection]
camlow325 has joined #jruby
andrewvc has joined #jruby
fidothe has quit [Ping timeout: 260 seconds]
prasunanand has quit [Remote host closed the connection]
fidothe has joined #jruby
asarih has quit [Ping timeout: 260 seconds]
asarih has joined #jruby
vtunka has quit [Ping timeout: 252 seconds]
<nirvdrum>
headius: Did you see that jnr-ffi issue?
cthulhua has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
temporalfox has joined #jruby
<enebo>
chrisseaton: eregon nirvdrum we are planning on 9.1.6.0 this week (conference release to sync some rails 5 fixes). Anything you want to get onto master atm?
<chrisseaton>
enebo: no thanks, what's on there is stable and works with a binary GraalVM
<enebo>
chrisseaton: great
<chrisseaton>
Do you know about these alias warnings?
<enebo>
chrisseaton: no did a commit remove them recently?
<enebo>
chrisseaton: yeah I remember this callconfig going away but not that it also removed these warnings
<enebo>
chrisseaton: I guess I would wait on an answer from headius on this. The commenting out part is weird to me. the warning itself was basedo n looking at callconfig so it sort of makes sense that the warning cannot be in that location (or al least with this code) but I don’t know why the code was left in commented
tcrawley is now known as tcrawley-away
pil-afk is now known as pilhuhn
riddley has joined #jruby
shellac has quit [Quit: Computer has gone to sleep.]
<GitHub119>
[jruby] kares created test-jossl-0.9.18 (+4 new commits): https://git.io/vX0Jr
<GitHub119>
jruby/test-jossl-0.9.18 3d8babd kares: upgrade to jruby-openssl 0.9.18 (staging)
<GitHub119>
jruby/test-jossl-0.9.18 c5ed103 kares: do preffered JCE unrestriction on Java 9 - set "crypto.policy"...
<riddley>
I'm having trouble with character encodings using Jruby 9.1.5.0 reading/writing postgresql... I'm building a hash that I hand off to Sequel. I check the encodings of the values and keys prior to trying to insert them and (for several problematic records) they are all UTF-8. But then jruby throws this error: WARN: Encoding::CompatibilityError: incompatible character encodings: ASCII-8BIT and UTF-8 (next line) WARN: org/jruby/RubyString.java:3145:
<riddley>
in `[]='
<GitHub100>
[jruby] kares opened pull request #4268: upgrade jruby-openssl to 0.9.18 (master...test-jossl-0.9.18) https://git.io/vX0J9
<riddley>
thousands of records are working fine, but a small few of them have this error
<riddley>
the keys/values also respond with true to valid_encoding?
<enebo>
riddley: I wonder if it is making a query and some bits in the query somehow thing it is binary (ascii-8bit)
<enebo>
riddley: if all your inputs are UTF-8 then something in sequel itself is making an 8bit string somewhere
<riddley>
ok ty
<riddley>
I'm reading from on DB, doing an ETL and then trying to insert into a new DB when I get that error
<enebo>
riddley: and the data you get back displays UTF-8 before reinserting it?
<riddley>
yea I loop over all of the keys and values in my hash and they're all UTF-8 and valid_encoding? == true
<riddley>
hundreds of thousands of records work just fine... a few hundred get mad like this
pilhuhn has quit [Excess Flood]
<enebo>
riddley: but those few records have multi byte chars right?
<enebo>
riddley: if you can prove all your strings you are putting into sequel are utf-8 and valid then I guess maybe try and check internally in sequel for ascii-8bit
pilhuhn has joined #jruby
<enebo>
riddley: unless the backend you insert into is not configured for utdf-8. I am just guessing
<riddley>
at first glance a bulk of them have multi-byte chars, yes. Some multi-byte stuff is working fine though.
<riddley>
postgresql (destination) is set to utf-8
<riddley>
enebo, on another note, I'm using jruby to make reading from Kafka easier (I know nothing about Java) and I get complaints about a lack of configured log4j appenders. Reading the URL in the error didn't really help me to understand what to do... is it something the jruby-kafka maintainers should be doing or me?
<nirvdrum>
enebo: Not entirely related, but are you planning on jnr releases?
<enebo>
riddley: I am not sure about the url you mean offhand
<enebo>
nirvdrum: Is there something for jruby-proper we need from a release?
<enebo>
nirvdrum: I know we have a new platform potentially in jfffi?
<nirvdrum>
enebo: There were the Solaris constants. I don't know if headius had a chance to verify if that fixed other issues.
<enebo>
nirvdrum: well I am leery to change anything at this point since I want to release in next day or two
<enebo>
nirvdrum: except perhaps kares_ jossl update assuming he feels really good
<nirvdrum>
enebo: I'd really evaluate that jnr-ffi one.
m4rCsi_ has quit [Read error: Connection reset by peer]
<nirvdrum>
As far as I can tell, that was merged after 9.1.5.0, so you have no releases with it.
<enebo>
riddley: I guess I don’t know anything about this. It probably does not hurt to ask kafka maintainers. If they bundle stuff up into an app server or make their own standalone server then I would think they should provide something to eliminate this warning
<enebo>
nirvdrum: so our current jnr-ffi release has this merged or no?
<nirvdrum>
2.1.0 does, yes. 2.0.9 was what was last shipped, I believe.
<riddley>
enebo, tyvm :)
<enebo>
riddley: again I am just guessing on that one :)
<nirvdrum>
enebo: But relying on a HashMap ordering you have no control over will lead to heartache, as I ran into.
cprice_ has quit [Ping timeout: 252 seconds]
<enebo>
I see we depend on 2.1.0 so this must be in that release
<enebo>
nirvdrum: so I guess we should fix the ordering
cprice_ has joined #jruby
<nirvdrum>
I think just making it a LinkedHashMap is good enough here.
<enebo>
nirvdrum: if you fix it I will reelease it :)
<GitHub8>
[jruby] eregon pushed 2 new commits to truffle-head: https://git.io/vX03I
<nirvdrum>
enebo: Okay. It's just a mess because I'll need to update the jnr-posix POM and do another snapshot then update the test environment I was in.
<enebo>
nirvdrum: are you going to rubyconf?
<nirvdrum>
enebo: I think this only hasn't been that big of a problem because there's some special-case handling of libc. But if you need librt (e.g., on older glibc) you're susceptible to it.
<nirvdrum>
enebo: Unfortunately, not.
<enebo>
nirvdrum: bummer. we are hopelessly behind right now in our plans so I appreciate your extra help here :)
<nirvdrum>
chrisseaton will be there.
<nirvdrum>
I'll be online for moral support :-)
m4rCsi has joined #jruby
<enebo>
nirvdrum: …and guidance
<nirvdrum>
enebo: Heh. Is 1.7 really EOL at the end of the year?
<enebo>
nirvdrum: well it is the plan but we have not fully evaluated what issues are left
<enebo>
nirvdrum: regardless though even if we continue it then it will be much less supported and of course 2.x support will never be these so that will increasingly make it unattactive
<nirvdrum>
enebo: Yeah. It's just become really clear to me that a lot of your overhead is supporting a ridiculous number of platforms.
<nirvdrum>
And two major branches at that.
<enebo>
nirvdrum: yeah we definitely want to stop supporting 1.7
<nirvdrum>
I was looking at the Ruby runtime on Friday trying to figure out in which situation SecureRandom isn't available.
<enebo>
nirvdrum: this recent arjdbc work should end up reducing overhead of that a lot for Rails 5.1
<nirvdrum>
I have no idea. But there's code there to support it. I assume someone needs it.
<nirvdrum>
Sweet.
<enebo>
nirvdrum: most the local sqlite3 code is a complete EXACT copy of the body of the AR adapter code
<enebo>
nirvdrum: so once we can get them to change how they load that we can load their copy and remove ours
<enebo>
nirvdrum: but I am concerned about maintenance. If we only follow latest rails we may end up with lots of ar-jdbc releases
pawnbox has quit [Remote host closed the connection]
<enebo>
nirvdrum: with that said the support all versions at once made the code more and more unmaintainable over time
<chrisseaton>
Does it sound right to people that rendering a page in a real Rails app might involve 1.8k regular expression matches?
<enebo>
nirvdrum: so more simple small releases might be preferable to less but more complicated releases
<enebo>
wow
<enebo>
chrisseaton: maybe? Sounds like a lot but who knows
<chrisseaton>
And writing to the caller frame $~, $_ etc 11k times
<enebo>
chrisseaton: I would not be siurprised to see regexp matches per column value
<enebo>
chrisseaton: but even so that seems like a lot
<chrisseaton>
ruby -e 14 involves 600 re matches
<enebo>
chrisseaton: ruby noes
<chrisseaton>
and 2246 writes to the caller frame
<enebo>
chrisseaton: how many gems?
<chrisseaton>
250
<chrisseaton>
This is Discourse I'm running
<enebo>
2 per gem load maybe seems realistic
<enebo>
err not load
<enebo>
load of gemspec
<enebo>
chrisseaton: —no-rubty-gems or whatever that option is
<nirvdrum>
enebo: Granted I've done zero advocacy work on this, but I'm a bit surprised Rails hasn't adopted some sort of DBI that you could the adhere to.
<enebo>
chrisseaton: that will at least tell you how much is gems from -e
claudiuinberlin has quit []
<chrisseaton>
ruby -e 'gets' takes exactly one more caller frame write than ruby -e '14', so this seems to be working!
<enebo>
nirvdrum: headius and I have been frustrated that sequel is not just an adapter
<enebo>
nirvdrum: then all dbs sequel supported would work
donV has quit [Read error: Connection reset by peer]
donV has joined #jruby
donV has quit [Read error: Connection reset by peer]
donV has joined #jruby
tcrawley is now known as tcrawley-away
enebo has quit [Quit: enebo]
<thedarkone2>
chrisseaton: re regexps, yes, not surprised, most of it is prob html/html-attr escaping, on MRI you'd use c-ext gems for that (they seamlessly monkeypatch the necessary methods)
<chrisseaton>
thedarkone2: yes I found Alexey comparing each character against an re
<chrisseaton>
thedarkone2: are you surprised that $_ and $~ are written 10x as many times as re matches?
<thedarkone2>
chrisseaton: so on jruby+truffle you're essentially running ruby regexp code that no serious Rails runs in ruby
<chrisseaton>
thedarkone2: not sure what you mean?
<thedarkone2>
a normal Rails app uses gems that replace pure ruby regexp html sanitization with C code (not regexp based), so in reality you are seeing Ruby code is almost never run by MRI, so people don't care about it
<chrisseaton>
I'm running unmodified Discourse on MRI for this paper.
<thedarkone2>
sorry then
<chrisseaton>
This lexer using re is in SmartyPants
<thedarkone2>
so, my statement above "yes, not surprised, most of it is prob html/html-attr escaping" was wrong :)
dling has joined #jruby
olle has quit [Quit: olle]
donV has quit [Read error: Connection reset by peer]
donV has joined #jruby
donV has quit [Client Quit]
olle has joined #jruby
olle has quit [Quit: olle]
<GitHub90>
[jruby] eregon pushed 2 new commits to truffle-head: https://git.io/vX06p
<GitHub90>
jruby/truffle-head 07d3630 Benoit Daloze: Let KCode be independent of the Ruby runtime...
<GitHub90>
jruby/truffle-head 19ca0f8 Benoit Daloze: Allow creating a RubyInstanceConfig without loading the Ruby class...