Race condition with reset of _self._cmds in invoke(). about child-shell

[rannn505]

Hey @mattflix,

You are right about cleaning up the array of commands and in the next version I will clean it immediately after receiving the confirmation that it has passed to the PS treatment.

With regard to race condition, the library is based on promises and therefore, using the existing API, it is very easy to avoid the situation you described.

Thank you very much for turning my attention. 💯 👍

[mattflix]

Yes, it is not very "difficult" to avoid the race, but the race is created entirely by the design of the existing API.

Due the current API, every library client must:
a) first be aware that there may be a problem (which they may not be until strange behavior emerges, possibly after they have shipped their code),
b) debug and understand the problem, and
c) possibly write extra code to ensure that any group of addCommand() call(s) and the final invoke() call are always executed as an "atomic" operation.

So, my feedback is that, a slightly different API design (like the one I proposed) would eliminate the possibility for the race altogether, rendering moot concerns a), b), and c), making the library that much more robust and reliable.