Jonathan Blow

My experience is that people hate base level unix tools (and come on, there are like three dozen that count) because they vastly overestimate their ability to improve upon them, regardless of how buggy and clumsy they are. And I'm not even sure that they are.

It shouldn't be hard to improve on them. They are from the 1970s. If we don't have better ideas 50 years later, we suck real bad. I agree that many people *don't* improve on them, but that is a different story.

Well... Rust community is currently making a credible effort of doing that properly. Following them with great interest. I'd use make reimplemented in Rust in a heartbeat (if well done etc.)

Check out stuff like ripgrep and follow the trail if this is news to you.

I think it's a good idea to rewrite things so that they are known to be more-correct, but the RIIR movement is not going to make things that much simpler, because the core problem is not the actual source code.

Hmm, I thought your main objection was the stuff about "they are buggy and clumsy and have weird failure states". For me as a spectator, it seems like they're at least following the right trail in that regard...

Well that's one objection of many. But a big part of the bugginess is not the bugginess of the individual programs, it's that they are combined via loose coupling, communicating with non-statically-typed textual data.

man... I'm actually SUPER torn on that one. A lot of devs might benefit from not getting anything but passing textual data as it could help them to see the complexity they're piling on top of basic, textual data...