<rsim>
We have identified one very strange Java bug on one particular customer instance - java.lang.Double.new(675.8818 / 145).toString returns "*****************" and java.lang.Double.new(65.896 / 188).toString returns "0.****************"
<rsim>
I was looking at java.lang.Double and sun.misc.FloatingDecimal source code but could not identify when such string of asterisks could be returned
<rsim>
What we know about this customer instance is that they have Java SE Runtime Environment 1.8.0_181 on Windows Server 2012 R2 6.3, OS Architecture amd64
<rsim>
Tried to reproduce it on our test Windows virtual machines with the same Java version but for us toString returns correct double value string representations. Anyone else has seen something like this? Could not find anywhere any similar bug description.
kiwi_50 has joined #jruby
kiwi_50 has quit [Remote host closed the connection]
rsim has quit [Remote host closed the connection]
drbobbeaty has joined #jruby
rsim has joined #jruby
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<rsim>
One more thing - we also tested (675.8818 / 145).to_s for this customer (using JRuby 9.2.6.0) and also got this strange "*****************" result
shellac has joined #jruby
drbobbeaty has joined #jruby
rsim has quit [Remote host closed the connection]
lucasb has joined #jruby
subbu is now known as subbu|busy
<headius[m]>
Wow that's...weird
xardion has quit [Remote host closed the connection]
xardion has joined #jruby
shellac has quit [Ping timeout: 250 seconds]
subbu|busy is now known as subbu
<headius[m]>
kares: you around?
<headius[m]>
I found a bug in MathLinker...it tries to shove a java.lang.Double into a method handle expecting a RubyFloat
<headius[m]>
it's just missing the creation of the RubyFloat, but I think we could be doing a better job of keeping the float on hand in the call site
lucasb has quit [Quit: Connection closed for inactivity]
subbu is now known as subbu|lunch
subbu|lunch is now known as subbu
whitingjr has quit [Ping timeout: 246 seconds]
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
snickers has joined #jruby
snickers has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
whitingjr has joined #jruby
whitingjr has quit [Ping timeout: 248 seconds]
whitingjr has joined #jruby
drbobbeaty has joined #jruby
whitingjr has quit [Ping timeout: 272 seconds]
<headius[m]>
kares: I pushed a fix; the fixnum path was doing the right thing
<headius[m]>
Both of these probably could be made into CallSite subclasses that receive the literal value and cache an object for it, just in case
<headius[m]>
the bug was preventing us from running Mandelbrot, but no tests failed, so I think we need to add something to cover this case