Comment by torben-friis
2 days ago
Have there been any significant issues caused by the vibecoded translation?
To be clear, I'm not implying support for the merge. I am against this whole YOLO approach to engineering. Just curious how the switch is going since I haven't seen any news since the merge announcement.
IMO the source of the new code is less important than the sheer volume of it. Bun does not need to be entirely rewritten; certainly not over a period of a week, possibly not even over a period of a year. Stability is hard-fought and battle-tested. Everyone has a plan until they get punched in the face; and every repository has passing tests until it runs production code.
How many lines of translated code would have been acceptable? What about before AI all the machine translated projects that people used for years without a single complaint? The person who did this was the lead for Bun from the get go no? So it stands to reason that they are one of the highest authorities in Bun itself.
What is the essence of a program?
The name? No.
The author? No.
The test suite? No, testing is hard and most projects don't do enough of it.
The code? Not really, code can be rewritten.
The history of the program being executed by its users? Ding ding ding. Yes, it's this one.
Bun no longer exists. What exists now is a program that has the potential to be Bun but at this time is not Bun because it has no history of being executed as Bun.
If the rewrite has been done piece-by-piece or over a longer time period then this history could have been built up over time. As such we will have to wait a while for Bun to exist again.
5 replies →
> What about before AI all the machine translated projects that people used for years without a single complaint?
If by machine translation you mean something like transpiling, that's a technology that has been proven over decades, and the translators were written by hand with some attempt at formal correctness and guarantees. Translating with LLM is much newer and subject to the errors LLMs can create, such as hallucination. And I think a lot of people would still be nervous about translating an entire project of that many lines, even with the best pre-AI translators.
I agree, I'm just wondering whether the punches have come already. Honestly, the fact that they haven't is against my expectations, my guess was that we'd see a noisy one here a day after merging.
It's too early. It might be too early forever.
Epistemically: if it can be so easily vibecoded to rust, why can't it be vibecoded to be reparted? Isn't the great and Almight AI unable to parse and repair Zig? Identify it's weak points and route around it?