3 ms·

> this seems like such a cop out. its like saying "oh failure is a given, so don't even try to succeed". It is a pragmatic cop-out yes. I found it is easier t

> this seems like such a cop out. its like saying "oh failure is a given, so don't even try to succeed". It is a pragmatic cop-out yes. I found it is easier to make progress by assuming that you'll get to Turing Complete rather than investing in the time to avoid it. I found that the downsides of Turing Completeness are usually overhyped or primarily theoretical. > so just because Rust (and others) failed, doesn't mean its impossible. It definitely is not impossible. But I don't know if it is worth it. In the end I want fast compile times, type safety, easy to maintain code and good error messages. I do not care if the type system is Turing Complete.

38 3y ago
> In the end I want fast compile times, type safety, easy to maintain code and good error messages. I do not care if the type system is Turing Complete. I think the problem is some languages with Rust go too far with generics, which probably triggers the turing complete. for example, this is valid Rust code: let mut handles = Vec::new(); for i in 0..10 { let handle = do_work(i); handles.push(handle); } but you have to follow the code all the way to "do_work" before you ever find the type of anything. Go does not allow this. you need to either declare a concrete type: var handles []int or a explicit generic type: type slice[T any] []T var handles slice[int] I think Rust is lose lose, because the underlying type implementation is Turing complete, and I would argue the code is actually less readable because of the overuse of type inference.
- bhouston 3y ago
  > I think Rust is lose lose, because the underlying type implementation is Turing complete I am not a Rust coder so I can't really comment. Currently, I do TypeScript, which also has a Turing complete type system, and I love it. Of course all things in moderation. Even though I could make a completely obtuse type design for projects, I try to not write code that others can not understand.

38 3y ago

> In the end I want fast compile times, type safety, easy to maintain code and good error messages. I do not care if the type system is Turing Complete. I think the problem is some languages with Rust go too far with generics, which probably triggers the turing complete. for example, this is valid Rust code: let mut handles = Vec::new(); for i in 0..10 { let handle = do_work(i); handles.push(handle); } but you have to follow the code all the way to "do_work" before you ever find the type of anything. Go does not allow this. you need to either declare a concrete type: var handles []int or a explicit generic type: type slice[T any] []T var handles slice[int] I think Rust is lose lose, because the underlying type implementation is Turing complete, and I would argue the code is actually less readable because of the overuse of type inference.

bhouston 3y ago
> I think Rust is lose lose, because the underlying type implementation is Turing complete I am not a Rust coder so I can't really comment. Currently, I do TypeScript, which also has a Turing complete type system, and I love it. Of course all things in moderation. Even though I could make a completely obtuse type design for projects, I try to not write code that others can not understand.