I wonder if slow-to-compile languages, such as Rust, are going to have a structural disadvantage in the world we're entering.
When LLMs write code, there are typically several iterations before getting what you want. This favours prototyping languages.
When LLMs write code, there are typically several iterations before getting what you want. This favours prototyping languages.
Comments
Often times the runtime portion takes way longer as a result (at least from the stuff I have worked on) 😂
On the one hand, if you're expecting multiple iterations to get something working then you want those iterations to be fast.
On the other hand though, languages like Rust help eliminate entire categories of bugs so you would potentially need fewer iterations.
If those tools also actively avoid those bugs then you're just back to suffering from slow iterations. If they don't then fewer iterations will benefit more...
Automatic verification is the cheapest, so it makes sense to spend more time in this part of the cycle.
This tradeoff is already considered in ML loss functions
so i end up just abandoning it and just winging it
It’s usually much harder to improve fundamental flaws in the language.
It remains one of the best designed languages imo, balancing pragmatism with purity. Maybe I should pick it back up again.
* A debugger with breakpoints + REPL access, à la Ruby or Elixir
* An easy to use GUI library with native components
* Blog posts, articles, videos for dumb people explaining pragmatically how to do real world things.
But the rest is wonderful!
¯\_(ツ)_/¯
I never fully learned it but it became my benchmark for language design (still is). I’m sure you can imagine how disappointed I’ve been through years of js, go, c#.