If we're going to use the "just use GPT" response, you could ask ChatGPT for a systemd unit.

Also, this completely ignores the fact that GPT4 will often make up functions or make incorrect assumptions about how a language feature works (and I've seen it make several mistakes wrt shell quoting and the like), making it unusable for writing production code when you don't know the domain yourself. If you now have to debug the script, the use case of LLMs is gone.

I really cannot believe how much worse HN comments have got over the past few years, from an already pretty annoying point.

just...wow.

I don’t know dude, did you try it? I’ve done exactly this, have it do the drudgery of writing the unit, I am a sophisticated developer. I am just trying to share a different POV, that the controversy over so many tools is really about how weird or wonky they are to use. But if the interface for them becomes simple and expressive, suddenly those complaints don’t matter. It goes from “just” and “simply” to just and simply, which is incredible. And if you still don’t know what I’m talking about, if you still feel like downvoting or saying what the other commenters say…

Listen, it’s $20 to find out if what I’m saying is true. Just try it.

I guess a (yet another) noticeable downturn happened at the Reddit exodus.

> Ask ChatGPT4 for a bash script that writes the unit for you into the right place. Colossal amounts of drudgery, all suddenly gone.

Given that humans can't write reliable bash scripts that don't randomly delete things in your computer [1], why would you expect something generated by LLMs to not randomly delete things on your computer or worse?

If you have to inspect and debug whatever the AI generated, that is potentially a lot worse that just writing it yourself in the first place.

[1] https://github.com/MrMEEE/bumblebee-Old-and-abbandoned/issue...