Post by @alexchapman@vee.seedy.cc

@alexchapman@vee.seedy.cc

2026-09-28 22:19 UTC

@jzb@hachyderm.io I believe that KDE made the right decision where it doesn't matter how the code was made, whether it was from an LLM, manually written in Notepad++ or another editor, or even written with pen and paper and then scanned into the computer and then converted to the right file. As long as the quality of the code is good and not just a one shot and done type of thing, then the way it was generated is irrelevant. @dansup@mastodon.social

Replies (1)

@jamescooke@hachyderm.io 2026-09-28 23:53

@alexchapman@vee.seedy.cc @jzb@hachyderm.io @dansup@mastodon.social > it doesn't matter how the code was made Hi - just reading this discussion, I disagree on this point. As a reviewer, I want to know what decisions were made - sometimes line by line. Why is this a bool? Why was a thing chosen to be a function and stdlib not used? Why use a file not a buffer? When reviewing generated code, there were no decisions. Questions about design choices are often answered with LLM generated text which *looks* like reasoning, but isn’t. At the other end of the scale is TDD built code that’s gone through multiple loops of Red Green Refactor. Here asking questions brings back detailed answers about why certain things worked or didn’t. Why things were mocked. Why extra layers were added or removed. When we know about why code is the way it is, it increases quality and maintainability - that’s why we care about how it was made.

Open ##5411171