ATS-Postiats VS rust

This is one of my two favorite books in The Little ...er series. The other is The Rational Schemer. These are two of the most advanced books in the series.

The Little Typer provides an introduction to dependent types. These can by used to guarantee things like "applying 'concat' to a list of length X and list of length Y returns a list of X+Y". It is also possible, to some extent, to use dependent types to replace proof tools like Coq. Two interesting languages using dependent types are:

- Idris. This is basically "strict Haskell plus dependent types": https://www.idris-lang.org/)

- ATS. This is a complex systems-level language with dependent types: http://www.ats-lang.org/

The Rational Schemer shows how to build a Prolog-like logic language as a Scheme library. This is a very good introduction to logic programming and the implementation of backtracking and unification is fascinating.

This is an excellent series overall, but these two books are especially good for people who are interested in unusual programming language designs. I don't expect dependent types or logic programming to become widely-used in the next couple generations of mainstream languages, but they're still fascinating.

Not being ATS

> any safety checks put into the competing language will have a runtime cost, which often is unacceptable

This is completely wrong. The best counterexample is probably ATS http://www.ats-lang.org which is compatible with C, yet also features dependent types (allowing us to prove arbitrary statements about our programs, and check them at compile time) and linear type (allowing us to precisely track resource usage; similar to Rust)

A good example is http://ats-lang.sourceforge.net/DOCUMENT/ATS2CAIRO/HTML/c36.... which uses the Cairo graphics library, and ends with the following:

> It may seem that using cairo functions in ATS is nearly identical to using them in C (modulo syntatical difference). However, what happens at the level of typechecking in ATS is far more sophisticated than in C. In particular, linear types are assigned to cairo objects (such as contexts, surfaces, patterns, font faces, etc.) in ATS to allow them to be tracked statically, that is, at compile-time, preventing potential mismanagement of such objects. For instance, if the following line:

      val () = cairo_surface_destroy (sf) // a type error if omitted

For a low level language in which you actually need to prove that your code doesn't cause UB, see http://www.ats-lang.org/

Here's the homepage fo the language: http://www.ats-lang.org/. The trick to finding results about with google is to search "ATS programming language".

In SPARK's case, you have to state your invariants in even greater precision than in Rust, and naturally it has worse inference. That's okay, the same happens in a certain language with Atrocious Type Syntax.

Yes, things like mentioned in the post can be expressed and checked statically, as demonstrated by languages like Idris and ATS. ATS might be even more relevant as it's an imperative language too, it can get rather low-level (like talking about properties of C runtime functions) while proving required properties statically, and it includes a solver for certain amount of arithmetics so that you don't need to prove obvious mathematical identities to the compiler. http://www.ats-lang.org/

You can use Tensorflow for Rust to simplify that task and avoid pain with regex. Just have the right mindset.

What is the benefit of this compared to using bindings/a wrapper to Tensorflow, or other ML libraries written in C/C++, such as this community hosted project on tensorflow's github. If it's just for fun that is a valid enough reason imo, just curious since you describe it as a better Tensorflow because of the typing vs using the python wrapper, when there already exist ways to interact with tensorflow with both Rust and other statically typed languages, also including C++ (officially supported), C#, Haskell and Scala, as well as probably having bindings not mentioned on the documentation for more niche languages.

(3) You could use TensorFlow as your executor: https://github.com/tensorflow/rust

TensorFlow has language support for TypeScript well as Rust.

Thank you for the restponse! With tensorflow I am probably better of with something like; [tensorflow rust bindings](https://github.com/tensorflow/rust/tree/master/src). But I believe some useful extensions are still written in python for example; [TFDV](https://github.com/tensorflow/data-validation).. and how about scikit-learn or even something that is simpler like fb-prophet that is entirely written in python?

Tensorflow bindings exist, technically, but they're in a pretty rough state AFAIK.

tensorflow/rust (Rust): Rust language bindings for TensorFlow