ur5us has quit [Ping timeout: 244 seconds]
ur5us has joined #jruby
_whitelogger has joined #jruby
ur5us has quit [Ping timeout: 260 seconds]
jeremyevans has quit [Ping timeout: 240 seconds]
nirvdrum has joined #jruby
drbobbeaty has quit [Ping timeout: 256 seconds]
rebelwarrior[m] has quit [*.net *.split]
XavierNoriaGitte has quit [*.net *.split]
ruby[m] has quit [*.net *.split]
Iambchop has quit [*.net *.split]
lopex has quit [*.net *.split]
den_d has quit [*.net *.split]
satyanash has quit [*.net *.split]
lopex has joined #jruby
den_d has joined #jruby
Iambchop has joined #jruby
XavierNoriaGitte has joined #jruby
rebelwarrior[m] has joined #jruby
ruby[m] has joined #jruby
satyanash has joined #jruby
jeremyevans has joined #jruby
<headius[m]> hmm seems like an actual bug on jnr update branch
<headius[m]> nearly good though
<headius[m]> we should be pairing
<headius[m]> I just figured that out
<headius[m]> I was thinking we could just replace the use of SocketLevel with IPProto here too, what do you think?
<headius[m]> I have a separate commit on that branch that starts using the jnr constants for this file too
<headius[m]> hmm maybe not? SOL_SOCKET doesn't seem to have an equivalent?
<headius[m]> looks like the value corresponds to IPPROTO_ICMP
<headius[m]> this setsockopt needs to be redone anyway for e.g. TCP_CORK
<ahorek[m]> it's because we need special paths to emulate sockets on Java, there're many workarounds and corner cases. CRuby basically just do a native call and that's it https://github.com/ruby/ruby/blob/e4be2fda3dbbfdb1f2ace697c96cf6bdd7dfef21/ext/socket/basicsocket.c#L304
<headius[m]> yeah we need a better impl of this to map to Java options if possible and fall back on native setsockopt like my TCP_CORK PR
<ahorek[m]> how hard would be to port https://github.com/jruby/jruby/issues/6334 from truffleruby?
<headius[m]> I don't know how much additional work they've done beyond the original rubinius impl but there are just a few key things that need to be done
<headius[m]> mostly it's dealing with the sockaddr structs
<headius[m]> rbx and TR both have to be rebuilt for target platform, at which point they generate the struct layout
<headius[m]> we don't have a build step
<headius[m]> I will pull your patch into my branch
<ahorek[m]> yeah Socket::SOL_SOCKET is special, others like Socket::SOL_IP... are Linux specific, but they have equivalents Socket::IPPROTO_IP
<ahorek[m]> ok
<headius[m]> I cherry-picked your fix into https://github.com/jruby/jruby/pull/6388
subbu is now known as subbu|lunch
<headius[m]> if it's green now (worked for me on a VM) I'll spin all the JNR releases and we can work on next steps
<headius[m]> thank you for getting all these constants added
<headius[m]> if we can figure out the socket structures, we could offer FFI sockets as a real option in 9.3
<ahorek[m]> nice
subbu|lunch is now known as subbu
<headius[m]> ok I have a working maven config that compiles appropriately for Java 8... I will incorporate that into all the JNR builds and then I think we can release them
subbu is now known as subbu|away
<headius[m]> ok all jnr is updated to use the Java 9+ release flag, to avoid problems with ByteBuffer etc
<headius[m]> I am rerunning the update PR build and if it remains green I'll spin all the releases
_whitelogger has joined #jruby
subbu|away is now known as subbu