sidx64 has quit [Read error: Connection reset by peer]
sidx64 has joined #jruby
<kares>
enebo: sorry about that I am lost in a few things I am trying to finish off, scrolled back but I am not able to find what you asked
<kares>
could you maybe re-ask please
<kares>
regarding 'noisy' I recall I have looked into the default state - on production sql logging is usually off
<kares>
that hack that was in previously with type_casted_binds: nil ensured that no noise is generated along the way
<kares>
otherwise you re-map all binds -> type_casted_binds pretty much for nothing, its feels like we do not need type_casted binds duplication
<kares>
as far as I can tell its only used for logging, so I was looking into that since this is pretty much a new thing in 5
<kares>
lack of proper abstraction ... as usual with AR - so I am not sure how you want to reproduce that, if that's that
<kares>
if I managed to find time I'd like to re-look into that (once again) whether its worth passing a few Rails suite tests (which feel artificial) with the change instead of not doing work that does not need to happen
coyo has joined #jruby
claudiuinberlin has joined #jruby
sidx64_ has joined #jruby
sidx64 has quit [Read error: Connection reset by peer]
sidx64 has joined #jruby
sidx64_ has quit [Ping timeout: 256 seconds]
adam12 has quit [Ping timeout: 260 seconds]
adam_ has joined #jruby
shellac has joined #jruby
sidx64 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
coyo has quit [Quit: leaving]
adam_ has quit [Ping timeout: 240 seconds]
adam_ has joined #jruby
sidx64 has joined #jruby
sidx64 has quit [Client Quit]
adam_ has quit [Ping timeout: 248 seconds]
adam_ has joined #jruby
sidx64 has joined #jruby
shellac has quit [Quit: Computer has gone to sleep.]
shellac has joined #jruby
adam_ has quit [Ping timeout: 264 seconds]
adam- has joined #jruby
adam- has quit [Ping timeout: 256 seconds]
adam- has joined #jruby
adam_ has joined #jruby
adam- has quit [Ping timeout: 264 seconds]
sidx64 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
adam_ has quit [Ping timeout: 256 seconds]
adam_ has joined #jruby
sidx64 has joined #jruby
rdubya has joined #jruby
sidx64 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
adam_ has quit [Ping timeout: 246 seconds]
adam_ has joined #jruby
sidx64 has joined #jruby
GitHub72 has joined #jruby
GitHub72 has left #jruby [#jruby]
<GitHub72>
jcodings/master 588eedf lopex: move multibyte caseMap to MultiByteEncoding
<GitHub47>
jruby/master c2ca66e Charles Oliver Nutter: Fix remaining failures related to Kernel/Warning#warn.
<GitHub47>
jruby/master 29d0b9c Charles Oliver Nutter: More correct and slightly more efficient kwargs Struct initialize.
<GitHub47>
jruby/master 484d537 Charles Oliver Nutter: Tag a refine spec; we'll fix them all at once, soon.
adam_ has joined #jruby
adam- has quit [Read error: Connection reset by peer]
machty has quit []
machty has joined #jruby
electrical_ has quit []
electrical_ has joined #jruby
adam_ has quit [Ping timeout: 265 seconds]
adam_ has joined #jruby
adam_ has quit [Ping timeout: 256 seconds]
adam_ has joined #jruby
adam_ has quit [Ping timeout: 240 seconds]
adam- has joined #jruby
sidx64 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
qmx has quit []
qmx has joined #jruby
adam- has quit [Ping timeout: 248 seconds]
adam_ has joined #jruby
<enebo>
kares: rdubya: sorry we did not revert exception method yesterday. I have hoped to get 9.1.17.0 but we ran into issues. We will be circling back today though.
<kares>
howdie, what's that?
<enebo>
The other reason we have not reverted is I think we were swallowing more than we wanted to so I am hoping to see why things like this are happening a bit: PostgresqlArrayTest#test_datetime_with_timezone_awareness:Java::JavaLang::ClassCastException: org.jruby.RubyNil cannot be cast to org.jruby.RubyTime
<kares>
oh I realized the ARJDBC thing, np
<enebo>
kares: well 9.1.17.0 I will finish making sure we are cool but that will be Monday since Friday releases are bad usually
<kares>
have seen that case - just tag it away :)
<enebo>
kares: but why is it happening?
<kares>
esp. if it gets you to change native stuf around ... users will complain if its smt real-worldy
<kares>
really not sure why - maybe its legit
<kares>
too much stuff on PG front so did not look at all
<enebo>
well I think it nil is acceptable input we shoudl probably just remove raising an exception if we can since it is bad for perf if nothing else
<enebo>
I may just spend a little time looking at what conditions cause this
<kares>
okay but I feel like lot of the suite is really artificial code we do not need to care about
<enebo>
What we have probably leads to right behavior but it rubs we have a ClassCast
<enebo>
test_datetime_with_timezone awareness passes normally it is only our changing catches which makes it fail
<kares>
like how they test -> its only in Rails suite and more will be added ... in real-world no one does it
<enebo>
and I am not sure if it is a weird corner test or not
<kares>
well your call +1 to tagit fro me
<kares>
actually tag (almost) everything!
<enebo>
kares: actually this was array so maybe this fell out of some that last rdubya and I think he was on board tagging this one...I picked a bad example I think
<enebo>
I may still look at it a tiny bit since it caught my attention
<rdubya>
enebo: that error with the RubyNil that you mentioned is the one I was talking to you about where the object looked like an OID::Array on the ruby side but not the java side
<rdubya>
that should be fixed
<enebo>
rdubya: yeah it may be I went back one from your commit on travis to see Charlie's last commit
<enebo>
NoMethodError: private method `select' called for nil:NilClass
<enebo>
perhaps stuff like that is more what I mean
<enebo>
ReaperTest very well may be something we care less about but us trying to dispatch indicates it is at least worth examining why they think we should be selecting on something
<enebo>
From by broader perspective on long term maintenance I want to document every line where we need to deviate with a clear explanation on why
<enebo>
That exception method headius overrode catches 5-6 Exceptions which I cannot see anywhere in Rails code. They probably need to be caught but I have no reason why we need to and I think makes reasoning against each successive version of Rails more difficult
<enebo>
5.1+ is down to a single RuntimeError check in AR so there code keeps paring down.
<rdubya>
Are you referring to the out of bounds exceptions?
<enebo>
rdubya: well that is simple we override a bunch of stuff in AR which was catching ActiveModel::RangeError so we needed extra catch in our own version because of that overriding
<enebo>
rdubya: however in 5.1 RangeError was made to extend InvalidStatement so we can remove it for 51+
<rdubya>
ah
<enebo>
but we wanted to see what broke across adapters just using Rails version for 50 and we noticed some other things caught and they are worth looking at a short amount of time
<enebo>
50 we probably revert and for 51 maybe we start paring down that method if possible
<kares>
headius: enebo: going side-ways back to JRuby I stumpled upon a simple idea to delay Java ext loading (there was some discussion on this previously)
<kares>
only load extensions as the particular Java proxy class is defined (instead of ahead-of-time as is currently)
<kares>
this would allow to scale defining more Ruby Java 'enhancements' wout worrying much on startup + help startup as well (for cases such as when JI isn't used)
<kares>
is it legit, or do you think I am missing smt?
<kares>
TL-DR: require 'java' won't define java.util.List Ruby extensions until List.class is being used in the runtime
<enebo>
kares: you know for IR persistence I end up loading all the defniitions of methods but do not actually decode any of the bodies since they are mostly not called
<enebo>
kares: do you mean something like that or just not define anything until the type is referenced?
<kares>
second thingy
<kares>
not define java.util.List#[] in Ruby land until java.util.List.class is getting used (proxy-class is being created)
<kares>
its kind of a 'lazy' Java Integration
<enebo>
kares: yeah it maybe is fine...I wonder about concurrency but we probably have same problem now it is just likely if we really have a race it loads before code is typically called (e.g. people java_import early in script). You probably understand that angle though
<kares>
that part already has its edge cases covered fairly well -> since proxy classes might define in parallel
<enebo>
kares: ok yeah makes sense we would make efforts
<kares>
there's some clever synchronization from headius around ... so this would only mean defining the ext part within that loading
<enebo>
kares: lazy is great when it works
<enebo>
kares: so we can definitely try but headius may have more of an opinion.
<kares>
yy I'll wai for him ... I did like the old days of explicit require 'java'
<enebo>
kares: startup time is our bane of existence :)
<enebo>
kares: I did too...I still say it is a semantic requirement
<enebo>
not that anyone cares since it just works without it
<GitHub100>
jruby/master d12091f kares: [fix] handle Time#at with Rational without loosing bigint precision
<GitHub100>
jruby/master 65be3cc kares: save an array instance (being copied twice) on byte[] -> String
<GitHub100>
jruby/master db2a375 kares: [refactor] simplify canonicalize negation check - internal after all
shellac has quit [Quit: Computer has gone to sleep.]
Puffball has joined #jruby
subbu is now known as subbu|lunch
Puffball has quit [Remote host closed the connection]
subbu|lunch is now known as subbu
subbu is now known as subbu|away
claudiuinberlin has joined #jruby
<GitHub164>
[jruby] enebo pushed 1 new commit to bytelist_love: https://git.io/vpmMt
<GitHub164>
jruby/bytelist_love 00cecc9 Thomas E. Enebo: Merge branch 'master' into bytelist_love
sidx64 has joined #jruby
<GitHub118>
[jruby] bogdanvlviv opened issue #5144: `RUBY_ENGINE_VERSION` and `JRUBY_VERSION` return version with suffix `SNAPSHOT` instead of `dev` for a dev version https://git.io/vpm7k
subbu|away is now known as subbu
bbrowning is now known as bbrowning_away
sidx64 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]