So I am undecided if this sort of growth methodology has ever been utilized to such an excessive earlier than so I figured I might doc it. In a nutshell, it is kind of like test-driven triplet-programming growth.
Whereas speed-developing our alpha codebase, 4 of us sat round a desk within the workplace in Berlin. Three folks (Vitalik, Jeff and me) every coders of their very own clean-room implementation of the Ethereum protocol. The fourth was Christoph, our grasp of testing.
Our goal was to have three absolutely appropriate implementations in addition to an unambiguous specification by the tip of three days of considerable growth. Over distance, this course of usually takes a number of weeks.
This time we wanted to expedite it; our course of was fairly easy. First we focus on the assorted consensus-breaking modifications and formally describe them as greatest we are able to. Then, individually we every crack on coding up the modifications concurrently, popping our heads up about doable clarifications to the specs as wanted. In the meantime, Christoph devises and codes exams, populating the outcomes both manually or with the farthest-ahead of the implementations (C++, usually :-P).
After a milestone’s value of modifications are coded up and the exams written, every clean-room implementation is examined towards the widespread check knowledge that Christoph compiled. The place points are discovered, we debug in a gaggle. Up to now, this has proved to be an efficient manner of manufacturing well-tested code rapidly, and maybe extra importantly, in delivering clear unambiguous formal specs.
Are there any extra examples of such methods taken to the acute?
