<enebo[m]> headius: I will try and build it in the morning. I was going to say I was not sure if I had it installed but I definitely did it for the false-positive virusy thing
<headius[m]> ok
<headius[m]> if I am bored this evening I might try to get it set up myself
<headius[m]> I hate this project, needs to be replaced
lucasb has quit [Quit: Connection closed for inactivity]
_whitelogger has joined #jruby
henry_bone has quit [Quit: henry_bone]
ur5us has quit [Ping timeout: 240 seconds]
ur5us has joined #jruby
<rg_3[m]> > rg_3: I believe so but they're kind of low priority while we try to wrap up the JRuby 9.2.11 series. I believe enebo fixed at least the `c-call` issue for 9.2.11 but he'll have to confirm that
<rg_3[m]> Thanks! 🙂
ur5us has quit [Ping timeout: 256 seconds]
shellac has joined #jruby
drbobbeaty has quit [Ping timeout: 240 seconds]
drbobbeaty has joined #jruby
_whitelogger has joined #jruby
lucasb has joined #jruby
claudiuinberlin has joined #jruby
<headius[m]> g'day folks
<kares[m]> ahoj!
<kares[m]> enebo: that Range IRubyObject[] boxing thingy looked weird
<kares[m]> anyhow what did you mean by "just using java_import" ?
xardion has quit [Remote host closed the connection]
<headius[m]> what Range boxing thingy?
xardion has joined #jruby
<headius[m]> kares: I think that's not a problem of java_import but how constants are looked up
<kares[m]> oh really, hmm that might explain why I thought it worked some time ago ...
<headius[m]> that's basically what java_import is doing but constant searches are lexical first, and a block scope isn't a constant (cref) scope
<kares[m]> right, I posted a `Class.new {}` version there ... but there's still a (possibly different) issue
<kares[m]> yeah I extracted the wrong thingy ... will need to update the issue.
<headius[m]> ok
<kares[m]> the problem I think is java_import from an instance method
<kares[m]> it ends up putting a 'global' constant vs a class method puts it correctly ... which should be an easy fix
<kares[m]> but not for 9.2 as it might be considered breaking
<headius[m]> could be it's just using Object if called against a non-class?
<kares[m]> yy
<headius[m]> should that even be defined in an instance method?
<kares[m]> should not for sure but shit happens ...
<kares[m]> I mean there's a warning but I guess it got lost
<kares[m]> will need to think about this - does not strike me as that much of an issue now that I got it reduced
<kares[m]> there's 2 patterns here: either import globally or in an instance method ... than there's no wonder conflicts happen :)
<headius[m]> huh yeah, we do define it on Object
<headius[m]> a decade later and I realize it should be defined on Module and toplevel only
<headius[m]> and it does evals, omg
<headius[m]> I have not looked at this code in a long time apparently
<headius[m]> kares: yeah it's clear why this works the way you are seeing
<headius[m]> if self is module it does a class eval of this string that defines the constant
<headius[m]> class_eval is unnecessary but there it is
<headius[m]> if it's not a module it does an eval with the current binding, which is java_import inside Object, so it defines it in Object
<headius[m]> we need to deprecate java_import against a non-module
<headius[m]> I'm daring to patch this
<headius[m]> feels like brain surgery
claudiuinberlin has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<headius[m]> kares: I have a patch that passes and deprecates Object#java_import
<headius[m]> passes spec:ji that is
<enebo[m]> kares: That single arg path may still be forcing boxing and could doCall maybe
<enebo[m]> kares: it was totally just in a recursive loop until that commit though
<headius[m]> aha one failure in -Ptest
<headius[m]> something is using java_import in a peculiar way
<headius[m]> aha
<headius[m]> not passing block through the wrapper java_import defs
rusk` has quit [Remote host closed the connection]
<enebo[m]> yeah I was wondering since I do not think people had seen any notifications yet on talks
<enebo[m]> it was >1000 so we were not going to be able to do it anyways
<headius[m]> yeah
<headius[m]> gonna get so much done this spring
<rdubya[m]> 😢
<headius[m]> the timing wasn't great for me anyway with my kid's birthday the day before
<rdubya[m]> yeah, it was rather late this year
<headius[m]> enebo kares: requested review on https://github.com/jruby/jruby/pull/6116
<headius[m]> it's green now
<kares[m]> looking good, but will visibility stay the same?
<headius[m]> negotiable
<headius[m]> you have a specific concern?
<kares[m]> testing things out ...
<kares[m]> oh right, this should work as before
<kares[m]> why do we need to also add java_import on main?
<enebo[m]> please think of what will specifically change here :)
<kares[m]> (java_import was only on Object before)
<enebo[m]> Seems reasonable to me generally as we should have done it this way to begin with
<headius[m]> it doesn't appear any code we have uses java_import against a non-module, other than obvious top-level calls
<enebo[m]> is there anything else which got some behavior difference from being done via module_eval which won't happen now?
<enebo[m]> self is self so I would assume a new constant would go to same place
<headius[m]> well it will parse using ConstDecl etc where I'm using the reflective calls now
<headius[m]> this code should be pretty close to equivalent but the syntactic constructs probably have minor differences
<headius[m]> it may be faster without the evals but I have not tried to measure that
<enebo[m]> meh
<enebo[m]> I mean that is fine but I am hoping people are not doing millions of java_import
<enebo[m]> I say land it for 9.3 and hopefully something will shake out if we get it in
<enebo[m]> well hopefully nothing will shake out but the more time the better
<kares[m]> 👍️
<headius[m]> the eval cost is likely lost in the overhead of loading a Java class proxy
<headius[m]> but less eval is always good
<enebo[m]> yeah I just doubt we will see it in anything other than a microbench
<enebo[m]> yep
subbu is now known as subbu|lunch
<headius[m]> ok I'll merge
<headius[m]> kares: up to you if you want to see anything else for https://github.com/jruby/jruby/issues/6115
<headius[m]> I'm moving on to other things
<headius[m]> enebo: anything from virus people?
<headius[m]> I'd like to pop this launcher issue off my stack
<kares[m]> fine by me
sagax has quit [Read error: Connection reset by peer]
<enebo[m]> headius: nope
sagax has joined #jruby
<headius[m]> kares: perhaps 9.3 should change Object#java_import to import into the self class?
<headius[m]> I'd rather just get rid of it (via deprecation) but that's another option
<headius[m]> so it would work like assigning a constant in a method (if that weren't a syntax error...)
claudiuinberlin has joined #jruby
subbu|lunch is now known as subbu
ur5us has joined #jruby
claudiuinberlin has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
claudiuinberlin has joined #jruby
travis-ci has joined #jruby
<travis-ci> jruby/jruby (ruby-2.6:e1d88ee by Charles Oliver Nutter): The build failed. https://travis-ci.org/jruby/jruby/builds/660807617 [258 min 19 sec]
travis-ci has left #jruby [#jruby]
ur5us has quit [Ping timeout: 256 seconds]
henry_bone has joined #jruby
ur5us has joined #jruby
claudiuinberlin has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]