Comment by mapontosevenths
9 hours ago
> unless it's coming from a known and trusted developer.
That's exactly the sketchy part here. They turned down known, working and tested, code that came from a partner (bun) due to this policy. Code that 4x'd compile speed.
A general ban makes sense based on their rationalization ("contributor poker"[0]). A total and inflexible ban can lead to a worse outcome for everyone though.
If a senior, experienced, contributor vouches for the code it shouldn't matter if they hand crafted it on stone tablets, generated it with yarrow sticks, or used gpt-3.
> That's exactly the sketchy part here. They turned down known, working and tested, code that came from a partner (bun) due to this policy. Code that 4x'd compile speed.
No; they turned it down because the vibe-coded PR was crap.
> The rewritten type resolution semantics were designed to avoid these issues, but Bun’s Zig fork does not incorporate the changes (and has not otherwise solved the design problems), which means their parallelized semantic analysis implementation will exhibit non-deterministic behavior. That’s pretty much a non-starter for most serious developers: you don’t want your compilation to randomly fail with a nonsense error 30% of the time.
> If a senior, experienced, contributor vouches for the code it shouldn't matter if they hand crafted it on stone tablets, generated it with yarrow sticks, or used gpt-3.
The flip side of that is that if such a contributor vouches for code that turns out to be poor-quality, this should severely damage their reputation. I've found far too many "senior" developers will give AI a pass on poor coding practices.
https://news.ycombinator.com/item?id=47958209
A standout paragraph from that thread:
> Put more simply, we are going to make these enhancements, but hacking them in for a flashy headline isn’t a good outcome for our users. Instead we’re approaching the problem with the care it deserves, so that when we ultimately ship it, we don’t cause regressions.
These exact changes are already on the roadmap and Bun’s PR is rushing ahead.
Thanks. That explains away most of my concern.
Quite the contrary, Bun's developers don't even understand language spec. Their slop didn't use the same type resolution semantics as Zig, which makes their implementation exhibits non-deterministic behavior.