The way Kaocha errors out in the case of a syntax error (namespace 'some.ns.with.error' not found) seems like a bad regression, I'm pretty sure this hasn't always been the case. Anyone else notice this?
don't really have time to do a proper disect now, but surprised this hasn't properly been reported
this is the kind of stuff I would absolutely sweat on when I initially developed kaocha
@jyn514 I'm guessing this is why you ended up vibe coding your own solution? this is why it's important to actually report an issue before jumping to solutions. I had no idea what the actual issue was. It seems we should first figure out where this regression happened so you get a stacktrace at the point of the error, making it pretty can be done after that
it's been like that since i first started learning Kaocha over a year ago, so i didn't realize it was a regression
anyway, i just won't share the code next time
why not?
because it feels bad to be accused of vibe-coding instead of paying attention
I'm sorry, I wasn't aware of that distinction. You said the code was LLM generated, that's all I was referring to.
thank you
it's not a big deal, i'm just a bit prickly because there's a bunch of people in the rust community who are very hostile to LLMs
this is my position on AI coding, it's from last September so not super up to date, but it still largely stands. I do use it in some cases, I do also have concerns and skepticism https://arnebrasseur.net/2025-09-15-position-on-ai.html
I'm not a fan of the hostility in some/many corners of the FOSS world, even though I understand where it's coming from https://toot.cat/@plexus/116706960262413961
yea
and as a FOSS maintainer I have particular concerns because it upsets a delicate social contract between maintainers and outside contributors, I still haven't fully settled on what my "official" stance is regarding AI contributions to my projects
i actually made a policy for the rust project around AI that's about to land, you might be interested in skimming it https://github.com/jyn514/rust-forge/blob/llm-policy/src/policies/llm-usage.md