In that one, G. van Rossum drops in to make a suggestion in the comments ...
Both posts are worth the time.
P.S.
some people do not realize that a sizeable part of the core Python developers were/are C++ coders who wanted some sanity (or so I gathered about 8 years ago, when facing the same problem ;-).
He lost me at his extreme advocation for unit tests. I won't rehash the arguments against them here but the short story is that there is an additional burden in writing and maintaining them, and many of their benefits can be had through other means. I agree with the gist of the rest of his points, which were mostly about reducing complexity. Unfortunately, unit tests increase complexity. They aren't a free lunch.
I had the same reaction. Unit tests mean you have twice as much code to change when requirements change. And in my experience, whenever a unit test fails you have to spend a lot of time debugging the test itself to make sure the failure is even valid.
I think most developers agree that unit tests are a good thing(tm).
I do agree that the authors assertion that not writing tests come down to A) ignorance or B) laziness. There is, after all, a third option. In a startup environment, where you're trying to rush a product to market, unit tests are often just not feasible. It's not ignorance or laziness, it's lack of resources to actually do it.
