force GC after benchmarking

[matthewleon]

Since this already uses hrtime, limiting it to Node, perhaps it's worth forcing the GC after benching?

https://simonmcmanus.wordpress.com/2013/01/03/forcing-garbage-collection-with-node-js-and-v8/

This would stop garbage generated during one run of bench from "leaking" into the performance of the next.

[paf31]

Sounds like a good idea, but is there no way to do it without requiring the --expose-gc flag?

[paf31]

I guess we could just do something like global.gc && global.gc(); and then the user can turn that flag on if they need forced GCs.

Let's add it!

[matthewleon]

I've given it some thought and realized it makes sense to issue a GC before benching as well, but before timing starts. It probably makes sense to then incorporate the concluding forced GC in the time statistic.

[paf31]

I'd be fine with either - perhaps we should offer a sum type of modes?

[safareli]

What about login a warning if global.gc is undefined?

[JordanMartinez]

This was implemented in #3. I opened #21 to track the other doc concerns.