Travis: Add 2.4.0, use jruby-9.1.7.0; don't trap QUIT on JRuby since it's not supported

[olleolleolle]

This PR

• adds 2.4.0 to the build matrix
• updates to latest JRuby

[michaelklishin]

Looks like tests hang on JRuby now, presumably because SIGQUIT is not handled as it used to be.

[olleolleolle]

@michaelklishin That message about SIGQUIT has been in JRuby a long time. Should Hutch change its tune about that signal?

[michaelklishin]

@olleolleolle on JRuby only, I think so.

[olleolleolle]

I was able to get the hang on QUIT with "Randomized with seed 60614".

[michaelklishin]

@olleolleolle 👍 let's drop QUIT support on JRuby then.

[michaelklishin]

By the way, we can do it in this very PR.

[olleolleolle]

@michaelklishin I'll see if I can get a fix on the thing, so that we don't... mess things up.

[olleolleolle]

@michaelklishin I began by unwinding a loop, so I could be remove a weirdly-placed next.

[michaelklishin]

@olleolleolle I'm not sure what you mean. But thank you.

[olleolleolle]

@michaelklishin The next move would be to never mention the QUIT signal to JRuby in the implementation of Waiter, so that no warnings are spouted.

[michaelklishin]

Makes sense, thank you!

[olleolleolle]

Now, the test output of WaiterSpec on JRuby is

Hutch::Waiter
  ["TERM", "INT"]
    includes only things in Signal.list.keys
  .wait_until_signaled
    a TERM signal is received
      logs that hutch is stopping
    a INT signal is received
      logs that hutch is stopping

[olleolleolle]

@michaelklishin Can you kill this build? https://travis-ci.org/gocardless/hutch/jobs/193808578

[olleolleolle]

@michaelklishin I'm more interested in the next build, which included the "skip QUIT on JRuby".

[michaelklishin]

@olleolleolle thank you!

[olleolleolle]

@michaelklishin Happy weekend!