xardion has quit [Remote host closed the connection]
Osho has quit [Quit: Lost terminal]
xardion has joined #jruby
Guest74956 has quit [Remote host closed the connection]
me has joined #jruby
me is now known as Guest62291
<kares>
marvinTD: looks weird esp. that it only happens with Rails
<kares>
seems like it might be rspeck/mock + rails related would try loading up the relevant bits in isolation (could try loading up some parts of rails + rspec/rails on that Sinatra project to confirm its smt specific to that env)
<GitHub141>
[jruby] brometeo opened issue #4985: An object with defined singleton method and no external references is not GC from memory https://git.io/vN4Da
<chrisseaton>
kares: I'm not sure what you mean - what do you think the output of to_int should be?
<chrisseaton>
In those particular cases
drbobbeaty has joined #jruby
<chrisseaton>
Commented on your issue
shellac has joined #jruby
<lopex>
there alaso might be a strictfp issue
<chrisseaton>
lopex: 1) strictfp only applies to the x87 co-processor, which nobody uses any more, 2) there are no calculations in this example which strictfp would apply to, and 3) the rounding to an imprecise value in this example applies in the parser
ped has joined #jruby
ped has left #jruby [#jruby]
bbrowning_away is now known as bbrowning
<lopex>
whoops, I shouldnt have opened my mouth
<chrisseaton>
sorry didn't mean to put you down! just explaining the facts
<chrisseaton>
But strictfp is largely irrelevant these days when anyone brings it up
<chrisseaton>
I think they're considering removing it from the JVM spec even
<GitHub184>
jruby/master 4a34053 Thomas E. Enebo: Mild refactoring of invokers code (came from bytelist_love work). As I need...
cremes has joined #jruby
<GitHub116>
[jruby] enebo pushed 3 new commits to bytelist_love: https://git.io/vNB2B
<GitHub116>
jruby/bytelist_love 49b9caa Thomas E. Enebo: Merge branch 'master' into bytelist_love
<GitHub116>
jruby/bytelist_love 7004a17 Thomas E. Enebo: Made method storage backed by ByteList instead of String. I knew this day...
<GitHub116>
jruby/bytelist_love ebb0f60 Thomas E. Enebo: Merge branch 'master' into bytelist_love
claudiuinberlin has joined #jruby
<byteit101>
I'm having some issues with IRB inside a non-forked PTY (`PTY.open {|m, slave| ... }`). Does anyone know if irb's readline needs anything past `$stdin.reopen(slave.dup)` for it to work correctly? I get a double echo, and arrow keys don't work until I hit enter.
bbrowning is now known as bbrowning_away
neoice has quit [Read error: Connection reset by peer]
neoice has joined #jruby
subbu is now known as subbu|lunch
<lopex>
enebo: so shall we land it ?
<enebo>
lopex: land what?
<enebo>
lopex: bytelist_love? no not yet. I have a massive changeset coming
<lopex>
enebo: or at least rerun the ci with new joni
<lopex>
I've run mri:fullint locally
<enebo>
lopex: I guess I don't know what you are asking
<enebo>
and man is travis horribly unresponsive
<enebo>
That zip inflater error :(
<lopex>
where
<enebo>
lopex: I see it happening on CI recently
<enebo>
lopex: not all the time though
<enebo>
lopex: I don't know if someone changed that code but it must have an issue
<enebo>
lopex: I am not crazy but InvocationMethodFactory.java:getPreSignature() is seemingly making the same string for every thing invoker in our runtime
<enebo>
thing = all
cremes has joined #jruby
<lopex>
why would it ?
<enebo>
sig is basically a string builder and params is basically making a primitive array and it is called for every single arity of every call() invoker method generated
<enebo>
so 4500 native methods * 4 calls() == 17000 stringbuilders to make 1 string?
<enebo>
anyways I think this is just a simple oversight but I thought I might be missing something
Aethenelle has quit [Ping timeout: 265 seconds]
<enebo>
oh actually I take that back it is not a huge number
<enebo>
What I said is true but we do not have very many methods which require full framing
<enebo>
4 * full framed invokers
<lopex>
you mean you want ton constantize ?
<enebo>
lopex: those are constants in spirit but not in practice
<enebo>
HAHA ok so I did a count
<enebo>
14 total calls!
<lopex>
not in practice ?
<enebo>
lopex: My original question was it is 100% the same string every call so why build it every call
<enebo>
lopex: the result I found is that it only happens 14 times
<lopex>
so cache it in enum instance
<enebo>
or just a static field
<enebo>
but I will only eliminate 13 array creates and StringBuilder instances during our build process :)
<enebo>
but naively I was thinking wow that looks expensive
<enebo>
If I look at this code in 3 years I will end up having the same reaction though :)
<enebo>
since I cannot remember much for very long
<lopex>
add a comment :P
<enebo>
yeah well on bytelist_love I might replace the function since only 1 call config is even supporter for invokers and I want to change some signatures in ThreadContext
<chrisseaton>
kares: I'm not sure it's possible to tell that a float's result is imprecise - the information is just lost and unless you do it with more precision simultaneously you can't tell
<enebo>
keeping all those old signatures around in case we want to play with different call configs seems like a lot of work for nothing
<lopex>
FrameFullScopeFull ?
<chrisseaton>
That's what integers do - they do calculations with one extra bit of precision - the overflow bit - but that doesn't translate into floats
<enebo>
lopex: yeah we are full or nothing
<enebo>
we can probably add more later but I would rather re-add code than support a bunch of unused code
<enebo>
I do appreciate the notion but call me lazy
<enebo>
chrisseaton: I doubt that BigDecimal issue will work out
<enebo>
chrisseaton: or perhaps another thought is it does not really solve all that much by changing it to BigDecimal
<chrisseaton>
enebo: I think Perl 6 has BigDecimal by default, or maybe it's Rational that's by default
<enebo>
I am sure Perl 6 does :)
<enebo>
Perl 6 does all the things
<enebo>
I just figure unless we go into some domain where all numbers are assumed to be unlimited precision you still see inexact values
<chrisseaton>
Perl 6 is an interesting study - it's approach to things - more than one way to do it, make it like natural language, enjoy being quirky, just become really unfashionable in the first decade of the century
<enebo>
based on that you are not much better off than just acknowledging how computers store floating point appears to be a little off
cremes has quit [Quit: cremes]
<enebo>
chrisseaton: yeah I guess so. tbh it demonstrated that to me in the last century but that went back to Perl 4
<enebo>
I believe I even ran Perl 3 for a very short time but I was not really using it back then
<enebo>
I was an avid perl programmer during the 90s until Ruby in about 2001 or so
<enebo>
Perl6 definitely jumped the shark on MTOWTDI
<enebo>
so it is a much better example of what happens when you go too far
<enebo>
chrisseaton: since you mentioned it I also found that moment switching from C++ to Java
<enebo>
C++ allows you to change too much surface behavior. Overloading () or =?
<lopex>
and ++() and ++(int)
<enebo>
lopex: well that is probably better than overriding *() and &() :)
<lopex>
avaik overloading ->* was something special
<enebo>
I did just wish yesterday Java had hygenic macros simlar to Rusts
<chrisseaton>
People always say perl is like speech - 'just write it like you'd say it'. But nobody ever explained why that was a good thing...
<lopex>
and const_expr
<enebo>
chrisseaton: so people can disagree with your speecg :P
<enebo>
the analogy totally holds!
<lopex>
the jokes goes any gibberish is proper perl
<enebo>
I am an extreme case though. Languages like Closure teach you one way to do things then a month later say, oh yeah but we all do it like this in reality with this weird esoteric syntax
<enebo>
Clojure
<enebo>
lopex: my joke was uuencoded text will run something useful in Perl
<lopex>
a quine
<enebo>
I dislike special syntax generally
<enebo>
(After just saying he wanted hygenic macros!)
<enebo>
nonetheless in Java world they could be first class and work well in tooling
<enebo>
anyways a man can dream
<enebo>
allowing user-defined syntax while complaining about conventional macros in clojure definitely makes me a hypocrite
cremes has joined #jruby
<lopex>
it's not a new syntax
<lopex>
and you cant do own reader macros in clojure
<lopex>
or you now ?
<enebo>
lopex: no you can't although people work around it
shellac has quit [Quit: Computer has gone to sleep.]
cremes has quit [Quit: cremes]
olle has joined #jruby
bbrowning_away is now known as bbrowning
olle has quit [Quit: olle]
Thyr has joined #jruby
Thyr has quit [Client Quit]
<GitHub193>
[jruby] nightsurge opened issue #4986: IO::Read is changing encoding from ASCII-8BIT to UTF-8 https://git.io/vNBb7