<headius> Yeah I am in Minnesota in the US where we have -30C winters and 40C summers so we have to handle both
subbu has joined #jruby
subbu has quit [Client Quit]
drbobbeaty has quit [Ping timeout: 252 seconds]
drbobbeaty_ has joined #jruby
razetime has joined #jruby
razetime has quit [Ping timeout: 252 seconds]
razetime has joined #jruby
<headius> enebo: so that PR is green but rsim has not confirmed it and this issue has only been reported under an OSGi rig
<headius> it's safe to go forward with but I don't know for certain that it fixes his issue
razetime has quit [Ping timeout: 252 seconds]
<headius> I am bisecting this regression on master in registry.rb
<headius> wow compose key really screws up vbox
<headius> hmm unsurprising but not particularly helpful... seems like it regressed around the time we updated net-* libraries to use gems
<headius> looks like the only change in there was that it now requires resolv.rb so I think I can narrow this further
<headius> well this update seems to be the one that brought in the breaking changes... small changes to win32/registry.rb: 900c531bd9bc57fb012290b9775eec548cc384f2
<headius> I will play with those changes in a bit and see if I can get back to working (and figure out why these changes broke)
<headius> ok so not the greatest news
<headius> the basic problem is that win32/registry switched from using our win32 FFI bindings to using Fiddle, and our Fiddle is not working right mapping parameters to types during the call
<headius> so it is able to bind the functions now but does not know when registry.rb passes a Fixnum for a pointer that it should be treated as a pointer value
<headius> since FFI wants a Pointer instance
<headius> the big fix is fiddle
<headius> short term fix might be simply reverting registry.rb to back to using our FFI win32
<headius> in any case I'm not fixing it today and it only affects master on Windows... so we'll revisit this next week
<headius> enebo: I'm logging off for the weekend. That PR above could be merged any time, but maybe we wait a bit and see if rsim can respond. Other than that I think we're on track for an early week 9.3.7 release