So I don’t know if this type of development methodology has never been applied to such an extreme before, so I thought I would document it. In a word, it’s a bit like the development of triplet programming focused on tests.
While developing the speed of our alpha code base, we sat around a table at the Berlin office. Three people (Vitalik, Jeff and I) each of the coders of their own implementation of the Warden Salle of the Ethereum Protocol. The fourth was Christoph, our test master.
Our goal was to have three fully compatible implementations as well as unambiguous specification by the end of three days of substantial development. On the distance, this process normally takes a few weeks.
This time, we needed to experiment with it; Our process was quite simple. We first discuss various changes in consensus revolution and officially describe them as best we can. Then, individually, we each crack on the coding of modifications simultaneously, arising our heads on possible clarifications to specifications if necessary. Meanwhile, Christoph designs and codes the tests, fulfilling the results manually or with the most distant from implementations (C ++, generally: P).
Once the value of the changes in an important step and the tests are written, each implementation of the white room is tested in relation to the current test data that Christoph has compiled. When problems are found, we debag in a group. So far, this has proven to be an effective way to quickly produce well tested code, and perhaps more important, to provide clear formal specifications without ambiguity.
Are there other examples of such techniques taken to the extreme?
