4 ms·
I made it through the first half. Even though it claims to be all about issues with modern hybrid-OOP languages, it's pretty much the same false dichotomy of pu
by geoelectric 7y ago
I made it through the first half. Even though it claims to be all about issues with modern hybrid-OOP languages, it's pretty much the same false dichotomy of pure OOP vs. pure FP that you see nowadays as everyone's "Yet Another Considered Harmful" post.
I'm sure there are a few decent observations in there that might be novel, but it's not really worth plowing through predictable rants about mutable state, inheritance, tortured DI architecture, and software-is-not-animals to get to them.
He didn't really acknowledge many of the things one might do in modern OO design to address those things, like favoring delegates and interfaces over inheritance, favoring shallow class hierarchies, or not inverting control just to unit test.
This isn't surprising, since one of the core tenets he puts out early is that his issue is OO is a loaded gun that lets you shoot yourself in the foot despite best intentions, vs. (apparently) FP that always keeps you honest by limiting your options.
Most of this probably sounds familiar. Overall, it feels like it's been written by someone who hasn't been around the scene long enough to know exactly how well-trodden their criticisms are.