setenv crashes with UTF-8 string

[xRandomGuy]

When JRuby is reading all envs, passing UTF-8 string to setenv crashes with IllegalArgumentException.

java.lang.IllegalArgumentException
        at java.nio.Buffer.position(Buffer.java:236)
!       at jnr.ffi.provider.converters.CharSequenceParameterConverter.toNative(CharSequenceParameterConverter.java:125)
        at jnr.posix.LinuxLibC$jnr$ffi$0.setenv(Unknown Source)
        at jnr.posix.BaseNativePOSIX.setenv(BaseNativePOSIX.java:303)
        at jnr.posix.CheckedPOSIX.setenv(CheckedPOSIX.java:301)
        at jnr.posix.LazyPOSIX.setenv(LazyPOSIX.java:303)
!       at org.jruby.RubyGlobal$StringOnlyRubyHash.case_aware_op_aset(RubyGlobal.java:424)
        at org.jruby.RubyGlobal$CaseInsensitiveStringOnlyRubyHash.op_aset(RubyGlobal.java:335)
        at org.jruby.RubyGlobal$StringOnlyRubyHash.op_aset19(RubyGlobal.java:387)
        at org.jruby.RubyHash$28.visit(RubyHash.java:1793)
        at org.jruby.RubyHash.visitAll(RubyHash.java:636)
        at org.jruby.RubyHash.replaceCommon19(RubyHash.java:1837)
        at org.jruby.RubyHash.replace19(RubyHash.java:1791)
        at org.jruby.RubyHash$INVOKER$i$1$0$replace19.call(RubyHash$INVOKER$i$1$0$replace19.gen)
        at org.jruby.runtime.callsite.CachingCallSite.cacheAndCall(CachingCallSite.java:326)
        at org.jruby.runtime.callsite.CachingCallSite.call(CachingCallSite.java:170)
        at org.jruby.ast.CallOneArgNode.interpret(CallOneArgNode.java:57)
        at org.jruby.ast.NewlineNode.interpret(NewlineNode.java:105)

[enebo]

The other detail to this report is that the default Java charset was not UTF-8. So the Converter was trying to set the position too large since the string ended up having more bytes than expected.

[xRandomGuy]

My default java charset IS indeed UTF-8
Setting LANG=C MAKES IT PASS and 'my locale chars' are turning into '?' which is acceptable in this case.

[enebo]

yeah I misunderstood our irc conversation. I think perhaps the crash is because the string length is longer than average # of bytes now:

ByteBuffer byteBuffer = ByteBuffer.wrap(new byte[(int) (string.length() * encoder.averageBytesPerChar()) + 4]);

I suspect this error will go away if we use maxBytesPerChar. @wmeissner what do you think?

[headius]

This appears to work fine in 9k.

$ jruby -e "ENV['foo'] = 'føø'; system 'echo \$foo'"
føø