←back to thread

Type checking is a symptom, not a solution

(programmingsimplicity.substack.com)
67 points mpweiher | 1 comments | 05 Sep 25 05:56 UTC | HN request time: 0.207s | source
Show context
bsaul ◴[05 Sep 25 20:00 UTC] No.45142957[source]▶
lots of negativity here. I found that POV very interesting. I'll rephrase like this: programmers tend to focus on micro-issues, but with large codebase the main problems are finding the correct abstractions and designs to manage the complexity, which themselves aren't described at all by types (a perfectly type safe program can be an absolute buggy over-engineered mess).
replies(1): >>45143049 #
1. ndriscoll ◴[05 Sep 25 20:09 UTC] No.45143049[source]▶
Aren't those abstractions described by types though? e.g. take a "paradigm" like Map-Reduce. Very buzzwordy for disturbed processing 15 years ago. Basic gist is "notice your program is a monoid homomorphism and you see it is trivially parallelizable". Or has been observed before, half of the Design Patterns in Java are just trivial uses of function types.