Why AI is pushing developers toward typed languages(github.blog)
21 pointsby ingvea month ago3 comments
  • aitchnyua month ago
    In my Django+Vue/TS/Inertia side project, I was surprised to see my agent know to build (use ts compiler in this context) after each change and it iterates till it gets it right. The Django code is not as typed, so I have to feed it a few error messages myself. Gotta unbreak my Mypy (python type checker) to keep my sanity.

    In 2012, I felt the high of Scala programs working perfectly once it compiles. Now my TS code is almost there, and Django is somewhat behind.

  • Havoca month ago
    Been vibecoding in rust for this reason. Even with the smaller amount of training data it does seem to produce less fragile code.
    • 9rxa month ago
      Why Rust, which is just barely an improvement over a dynamically-typed language, when you can just as easily vibe to a language with a proper dependent type system?
      • mips_avatara month ago
        I think a big part of why rust vibe codes so well is the care the rust developers put into their error messages. It helps the ai a lot
      • indemnitya month ago
        Which languages would that be? Not being snarky, actually curious.
        • 9rxa month ago
          Rocq or another in that wheelhouse.

          Once you go down the spectrum then you have to start relying on tests, which end up indirectly validating the same things as the type system anyway, at which point you are no further ahead than having used a dynamically-typed language.

          Partial statically-typed languages like Rust are pretty cool for things like highlighting basic mistakes in your editor as you type, automatic refactoring, etc. That is all very useful for humans writing code, even if a bit redundant on some technical level. But if an LLM is doing the work, none of that matters.

          • Havoca month ago
            Are LLMs any good at generating rocq code?
            • 9rxa month ago
              I asked several LLMs on your behalf and they all said "yes". But they might be a little biased.

              Why not put them to the test? It is only a prompt away. Rust is an available extraction target for Rocq, so you can even integrate it into your existing codebase.

              • Havoca month ago
                >Why not put them to the test?

                haha I shall. I like the idea of tighter constraints on LLM hence my enthusiasm for rust over python for vibecoding.

                I did just try it and the first thing it did was try to download half a gig. Which is fine, except my fiber install is next week and I'm hotspotting off a phone. So that'll need to wait

                ...it did seem to fundamentally understand the ask though of "make me a plan to make a http hello world server".

                Though at same time flagged that this isn't ideal

                >Rocq/Coq is primarily for theorem proving; HTTP server implementation will be minimal

                • 9rxa month ago
                  > Though at same time flagged that this isn't ideal

                  I suppose that depends how deep down the rabbit hole you want to go. HTTP is pretty minimal in and of itself. The complexity is due to it being layered on top of many other things, like TCP/IP, sockets, network devices, etc. Is it worth proving that your NIC functions according to spec? Probably not. But, if an HTTP server is what you are trying to build, like, parsing an HTTP request is quite suitable.

                  But a good case. I would suggest that Rocq is already proving itself. I expect by "make a http hello world server" you don't mean: First, invent the universe. Although maybe you do! Who knows? Your prompt is not clear. And since the type system is comprehensive enough to force you to actually think about what you really want, it is compelling you to be much more specific.

                  • Havoca month ago
                    >I suppose that depends how deep down the rabbit hole you want to go.

                    I'm intellectually curious about whether a formal verification style vibecoding would work. It seems like the logical outcome of current trajectory - if compute increases exponentially then something where the LLM can just hammer away until it checks out seems ideal way of vibecoding while getting a bullet proof outcome.

                    Knowing absolutely nothing about formal verification I'm quite skeptical it's viable right now though. (For hobby projects, I'm sure the formal verification academics are having fun)

                    So still leaning towards rust as current optimal. I am ofc biased (python/rust is my preference depending on task).

                    >I expect by "make a http hello world server" you don't mean

                    There is no deeper thinking behind it. Selected it purely because if I ask a LLM to oneshot something without guidance I think keeping it simple and within training data is fair.

                    • 9rxa month ago
                      > Knowing absolutely nothing about formal verification

                      Are you sure about that? That is what we've been talking about all along. After all, the thing that separates Rust from Python, with respect to our discussion here — the thing that allows code to be less fragile, is its type system. Which is formal verification! Rust is very limited in that regard, not allowing verification of much, but still very much lives on the formal verification spectrum.

                      The trouble when you can't verify much is that you end up in a situation where something like:

                          fn add(a: i32, b: i32) -> i32 {
        a + b + 1
    }
                      Is able to compile just fine despite being completely wrong. To deal with that you need to write tests (which is essentially runtime formal verification). But once you've written the tests, you've also largely covered the properties that are statically verified, leaving no real benefit to LLMs.

                      But I infer that you aren't writing tests either, and that Rust's type system is feeling advantageous as it at least provides some limited verification. You seem to be enjoying that workflow, which is why I mentioned that there are languages with way better types systems that you might enjoy even more. There are probably hundreds of languages between Rust and full verification on the spectrum, so if Rocq isn't the exact one for you, try some others! They're only a prompt away.

  • empthoughta month ago
    Python is at least as typed as Lua.