<headius>
eregon: oops didn't see this notification, I'll do that
<headius>
eregon: I'm just going to rewrite how we do globals...I can't refit that stupid API
<headius>
I plug one hole and another appears. It was never designed for concurrency and there's dozens of races
<headius>
probably never found because people rarely update globals, especially across threads
chrisarcand has joined #jruby
akp has joined #jruby
subbu is now known as subbu|lunch
claudiuinberlin has joined #jruby
<headius>
enebo: hey, in process of this global thing and wondering if I should key it off symbol or bytelist rather than string
<headius>
I don't know if we have any reported bugs about m17n global names but it probably doesn't work right
<enebo>
headius: yeah well symbol_love is trying to use RubySymbol but one thing that has me worried about the idea is constants
<headius>
worried why?
<enebo>
headius: bootstrapping
<enebo>
I did manage to get it working for methods so it is probably ok but constant registration is before method definitions
<enebo>
for Object/Class/Module
<enebo>
although I think it is deferrable
<enebo>
but that is the main question
<enebo>
anyways symbol_love will be massive codebase change so many things will change so we can just change again if need be to globals infra
<headius>
ah, sure I see what you mean
<enebo>
headius: but bytelist is a sure thing because it has no bootstrapping issues
<headius>
keying off bytelist is simple enough
<enebo>
the more I think about that work the more I feel like bytelist would solve the same issues except for it would still maintain the Symbol table stuff similarly to today
<enebo>
So RubySymbol is more appealing
<enebo>
but most of it is achieved with ByteList and it is less complicated during the genesis of our runtime
claudiuinberlin has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<headius>
it will try to run ant quickly to get environment details when you require it
<headius>
pretty awful error thought
<chrisarcand>
This mind sound silly but do I need to install ant separately? I guess I was expecting it to yell at me appropriately if that's the case
<headius>
you do and it is silly (silly that we don't ship it)
<headius>
I vote we move ant integration to a gem that pulls ant from maven
<chrisarcand>
^ That's partially why I'm so confused, I searched expecting a gem that maybe wasn't added to the project's gemfile but no such ant gem exists it seems.
<chrisarcand>
Aye, works great with ant installed and happy. thanks headius
<chrisarcand>
headius: Should I open an issue somewhere detailing the ant thing...? Would love to see that sort of unexpected awkward barrier removed from people trying out JRuby :)
<headius>
yeah do that
<headius>
moving to a gem would be trivial and if we use ant integration at all in our build it should be easy to change
claudiuinberlin has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
claudiuinberlin has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
akp has joined #jruby
claudiuinberlin has joined #jruby
claudiuinberlin has quit [Client Quit]
olle has joined #jruby
claudiuinberlin has joined #jruby
claudiuinberlin has quit [Client Quit]
akp has quit [Remote host closed the connection]
akp has joined #jruby
akp has quit [Remote host closed the connection]
<GitHub9>
[jruby] olleolleolle opened pull request #4806: Add Ruby part of TracePoint impl (master...feature/4800-tracepoint-trace) https://git.io/vdRdG