Restart the old-fashioned way

[macintux]

Related to #131, and per discussions with cliserv, restart isn't all that reliable. Here's a semantic change that may or may not be the way we want to proceed, but since I was in the mood to do something about it, here it be.

Worked "properly" (per the new semantic) in simple testing.

/cc @jaredmorrow @lukebakken

[macintux]

It occurred to me later that I didn't attempt to handle the case where stop failed (by not attempting a start); on the other hand, I'm not certain, that's the right thing to do. Maybe blindly trying to start it is best in case it does in fact stop.

[reiddraper]

Is this new code, or a 'revert' to the way it used to work? If so, what's the SHA of the old code?

[macintux]

Maybe @jaredmorrow knows otherwise, but I'm not aware of any previous incarnation of this behavior.

[reiddraper]

Ah, when I read 'the old-fashioned way', I was thinking that meant the way 'it used to work'. Carry on.

[macintux]

Gotcha. No, just John trying to be too clever. "The blue collar way"

[reiddraper]

Ha, fair enough.

[reiddraper]

Testing with Riak now. @macintux what do you think about amending the commit to include a more detailed commit message? And maybe a comment about restart continuing to try and start the node, even if the stop failed?

[reiddraper]

I've done the following to review:

• Read through the code
• Test this branch of node_package with Riak
• Check for bashisms (checkbashisms priv/base/runner)

+1 from me. Might be wise to have @jaredmorrow take a look as well though.

[macintux]

I'll update the commit message

[jaredmorrow]

+1

[macintux]

All cleaned up

[jaredmorrow]

👍