He talks about sending a browser a ton of random flexbox cases and determining where the elements ended up, and ensuring that his implementation ended up doing the same thing.
CORALINE: One of the things I've done as well, and this is a weird testing technique, from changing the implementation of an algorithm, I'll actually capture the old algorithm inside of a lambda in my test suite. I will throw values at the old algorithm by invoking the lambda and capturing the result, and then call the new function and compare the results to make sure that I haven't inadvertently changed something about the implementation that's changing the output characteristics of a method. And again, that's not something I would want to leave in place for very long but it's just a way of double-checking that you haven't changed something unexpectedly.
AVDI: That's cool. That's like the thing at a much coarser scale, larger scale, where you have two separate systems and you apply the same requests to both of them and log the responses and make sure that they match.
blog-post-move-fast-and-fix-things
talk-easy-rewrites-with-ruby-and-science