lucasb has quit [Quit: Connection closed for inactivity]
_whitelogger has joined #jruby
subbu|away is now known as subbu
nirvdrum_ has quit [Ping timeout: 244 seconds]
byteflam1 has joined #jruby
<kares[m]> headius: the caching site thing should work, regarding :
<kares[m]> > all it's checking is that it's a FixnumEntry, but it still could have been overwritten and that would not be seen, right?
<kares[m]> the FixnumEntry is only set (setCache override) for a RubyFixnum target (and is it a builtin) otherwise its a plain old CacheEntry
<kares[m]> hmm but I now get the point when + get redefined the cache won't notice since there isn't a cache token check
byteflam1 has quit [Ping timeout: 244 seconds]
whitingjr has joined #jruby
<kares[m]> yeah, definitely missed a `typeOk` check, adding those its still comparable to 9.2.7
<kares[m]> around 5% slower no accounting for statistical errors
shellac has joined #jruby
drbobbeaty has joined #jruby
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
whitingjr has quit [Quit: Leaving.]
<kares[m]> headius: updated the PR, could be simplified a bit but not sure its worth the hustle
<kares[m]> let me know if you feel like its still not usable ...
_whitelogger has joined #jruby
drbobbeaty has joined #jruby
nirvdrum_ has joined #jruby
lucasb has joined #jruby
<kares[m]> hmm, I've been lookin at more enumerator work -> getting compatibility improved
<kares[m]> we're missing support for feeding a return value, but when I add it perf. degrades
<kares[m]> to check the feed-ed value after yielding to the outside block
<kares[m]> the notable red flag is coming from a regression spec: https://github.com/jruby/jruby/blob/master/spec/regression/GH-1076_enumerator_finalization_needs_a_spec.rb#L19-L24 ... than runs out of memory
<kares[m]> seems that we won't be able to support enum.feed is we need to keep the enumerator instance completely separate from the fiber
<headius[m]> The Fiber can reach the Enumerator via weak references
rusk has quit [Remote host closed the connection]
Ryctolagus has joined #jruby
<headius[m]> ok I can look now
<headius[m]> 🤓
xardion has quit [Remote host closed the connection]
xardion has joined #jruby
<headius[m]> kares: I'm confused, can you show me a patch?
<headius[m]> kares: I think having the state object and the enumerator hold a single-element feed value array would work ok
shellac has quit [Ping timeout: 250 seconds]
<kares[m]> okay, that sounds simple enough
<kares[m]> will give it a try later, thanks
byteflam1 has joined #jruby
byteflam1 has quit [Ping timeout: 244 seconds]
byteflam1 has joined #jruby
byteflam1 has quit [Ping timeout: 246 seconds]
byteflam1 has joined #jruby
byteflam1 has quit [Ping timeout: 272 seconds]
byteflam1 has joined #jruby
byteflam1 has quit [Ping timeout: 272 seconds]
byteflam1 has joined #jruby
byteflam1 has quit [Ping timeout: 272 seconds]
byteflam1 has joined #jruby
enebo has quit [Ping timeout: 245 seconds]
knu has quit [Quit: Reboot...]
knu has joined #jruby
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<kares[m]> it worked, the code is a bit uglier and there's some duplication but I'll try putting it up to shape
<kares[m]> heh, lovely on every boot RGs seems to spin a few enumerators (over arrays) just to parse gem versions
byteflam1 has quit [Ping timeout: 268 seconds]
<headius[m]> of course
<headius[m]> we need to get more of the fast-path generator logic in there for core types
<headius[m]> since most of the time it's just eaching
<headius[m]> kares: you said you might look at this... https://github.com/jruby/jruby/issues/5284
<headius[m]> enebo and I are trying to scrub 9.2.8 issues down for release
<kares[m]> wasn't there a fast version for RubyArrays?
<headius[m]> In the Java impl...that did not make the migration to a fiber-based system but it's not hard to add back
<headius[m]> I have no numbers to know whether it helped but we should probably at least make it on par with 9.2.7 in that regard
<headius[m]> the logic is just defining to_generator on whatever the type is and producing a non-fiber generator
<headius[m]> add as many as you like :-)
byteflam1 has joined #jruby
<lopex> headius[m]: what's the state of jruby's docker support ?
<lopex> or how many contenders there are
<headius[m]> well the "official" one is the one in that issue
<headius[m]> we are supposed to take it over
<headius[m]> or rather you're supposed to take it over 😀
<headius[m]> it's all there, we just need to take ownership
<lopex> headius[m]: yeah, I said explicitly about my volunteership before
<headius[m]> ping that issue and cpuguy will hook you up with anything you need
<headius[m]> I will monitor so I can also be an owner of the image
<lopex> done
<headius[m]> 👍
<lopex> it's never ending
<headius[m]> Haha yes
<lopex> "The only people who might want /[a-z]/i to match U+017F are people handling Fraktur and Gaelic languages"
<lopex> how much fun can it get
<lopex> and I thought ascii only \w was hard
<lopex> it resuired new opcodes :P
<lopex> *required
<lopex> headius[m]: will you ever have a time to profile that onigmo cext against joni and I can help with that ?
byteflam1 has quit [Ping timeout: 246 seconds]
drbobbeaty has joined #jruby
nirvdrum_ has quit [Ping timeout: 272 seconds]
lucasb has quit [Quit: Connection closed for inactivity]