Hahahaha yeah, makes sense. Every time I use claude after codex I feel like I’m holding Claude’s hand the entire time. I imagine they have a lot of containing they have to do internally
878 karma · joined December 7, 2012
So, It’s neat to see something competent! Imagine if they modeled what cutting off the natural draining to the Everglades would do :p
https://www.anthropic.com/research/2028-ai-leadership
They are already starting that now.
There ya go, the rewrite was for marketing.
They’re trying to scale from 1 billion commits last year to over 14 billion this year. I have zero desire to try and manage that scaling. Basically being DDOS’d by agents all day now.
No healthy engineering team is going to do that. And I’d want to distance myself as far as I could from a project that behaves like that.
I won’t lie, if I had the access to this, I’d do the same exact thing.
> An example of this is the changes to type resolution which happened in the 0.16.0 release cycle—these didn’t affect users too much, but had big implications for the compiler implementation. Before those changes, the compiler’s behavior was often highly dependent on the order in which types and declarations were semantically analyzed by the compiler. Some orders might result in successful compilation, while others give compile errors. Single-threaded semantic analysis prevented these bugs from causing user-facing non-determinism. 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.
There is a reason why, zig is upholding the quality and they hate it.
I have a hard time believing they have something figured out that the rest of software engineers don’t. I’ve been using LLMs to help code since before claude code and heavily since. I have a hard enough time managing the “slop” from 4-5 agents. I can’t imagine this is going to go well having this done in like a week.
If you think that it will, I implore you to attempt the same in your large existing codebase and report back the mess. Theirs is littered with TODOs like this https://github.com/oven-sh/bun/pull/30412/changes#diff-6c2b5...
On top of that, every Tesla I have driven is poorly built. Ugly rattly plastic, bad panel gaps from the factory, factory paint not even matching, poor UX with everything jammed into the touch screen (lmao gotta go 4 levels deep in the UI to pop the glove box, unless they shipped a change for that). The brake pedal feel like stepping on a hard brick. No feel. Nothing. The drive by wire steering is like driving a 2005 nascar sim with no force feedback.
I could go on, they are objectively bad cars and for those who don't know cars and just want tech.
Cannot imagine this agent rewrite had anyone review any the code (you can’t at that speed).
I’m positive this will go extremely well :p
But you’re exactly right. They’re for the polished shoes folks, not the steel toes
It’s the most poorly engineered “truck” there is. Can’t tow. Can’t haul (stupid bed design). It’s just a glorified pavement machine.
This one is including external LLM services as apart of GitHub being “down”.
If you feel the need to do this, it’s time to move onto a tool you trust?
However you feel about OpenAI, at least their harness is actually open source and they don’t send lawyers after oss projects like opencode
Damage is done for me though. Even just one of these things (messing with adaptive thinking) is enough for me to not trust them anymore. And then their A/B testing this week on pricing.
I’ve been hacking on a macOS app that leans on LLMs, vision use, and the AX macOS APIs to try and make voiceover less.. prickly haha. Hoping to visit in person soon to watch her use it :)