14:12 <enebo[m]> ahorek: how are you benchmarking Rails?

14:13 <enebo[m]> ahorek: Small hashes could be from lots of kwarg calls perhaps. There is an activity for you if you are up for it

14:14 <enebo[m]> Although maybe we should figure out how you are getting your hotspot to see what the hash is

14:15 <enebo[m]> In Ruby 3 we will know if the hash is literal kwargs and so we realized we can make a LiteralHash in addition to Hash where the Literal Hash will just store the cached Hash rather than make it

14:16 <enebo[m]> This could still be helpful for Ruby 2.x but it would require that we internall call clone once. This may still be faster than running that helper (hard to say)

14:17 <enebo[m]> puritylake: I was on vacation last week

14:18 <enebo[m]> puritylake: I just restarted the last CI run and it appears to be updated

14:21 <puritylake[m]> Cool, rather sick atm so I’ll get back to you in a few days

14:27 <enebo[m]> puritylake: ok. get well soon

14:53 <headius> good morning

14:54 <headius> ahorek: yeah I am curious what you are benchmarking too... generally we should be at least on par with MRI

14:56 <headius> anything that much slower and I would assume we are failing to JIT some hot code

14:58 <enebo[m]> Or maybe some excessive exceptions

14:59 <enebo[m]> we were not even 8x slower cold last time I benchmarked

14:59 <headius> Yeah another one to look for

14:59 <headius> Right, seems like something is broken

15:32 <headius> enebo: so welcome back, what's on your plate this week

15:35 <enebo[m]> headius: what are you working on? I am looking at 3 spec runs

15:35 <enebo[m]> I restarted your last CI job this morning so the jnr stuff was updated

15:51 <headius> Well I think we need to prioritize getting the keyword argument stuff working so I will probably spend time on that. I think my Christmas project will be finishing m1 support in JFFI

15:52 <headius> The JNR releases took a really long time to propagate last week, I'm not sure why

15:53 <enebo[m]> Some recent changes broke what I am guessing is zsuper

15:54 <enebo[m]> looks like Symbol.getBytes -> null

16:03 <headius> Hmm

16:04 <enebo[m]> spec:compiler hits it

16:04 <enebo[m]> KW_REST_ARG_DUMMY

16:04 <enebo[m]> very likely seeing this

16:09 <headius> I haven't merged my kwarg changes yet

16:09 <headius> They won't fix as much as they break though right now

16:11 <enebo[m]> ok

16:20 <headius> Probably need to take a step back and isolate what we have for Ruby 2 compatibility and duplicate that logic to be modified for Ruby 3

16:58 <edipofederle[m]> Hi, quick help,

16:58 <edipofederle[m]> At `RubyHash` I'm trying to add a new entry to the hash with:

16:58 <edipofederle[m]> The think is the value should be a IRubyObject, there is a way to convert any type (string, int, long, etc) to a IRubyObject? or I'm looking into wrong direction here?

16:58 <edipofederle[m]> `op_aset(context, newKeySym, value);`

17:07 <enebo[m]> edipo.federle: JavaUtil.convertJavaToRuby

17:08 <enebo[m]> If you know then you should just do the right thing (aka runtime.getFixnum(int))

17:10 <edipofederle[m]> enebo: I see, I was looking into this one, let me try

17:11 <enebo[m]> All java objects when we script Java as Ruby uses this method

17:11 <enebo[m]> for all returning java objects

17:12 <enebo[m]> So it does more than just find the right core type

17:18 <edipofederle[m]> great. Thx!

21:45 <enebo[m]> headius: You do not have permission to push to this gem. Ask an owner to add you with: gem owner rake-ant --add tom.enebo@gmail.com

22:13 <headius> enebo: picking up kid from school, I'll get you it in there in a sec