Compiler cannot infer if an object of `impl <trait>` has impled another trait in Rust

Question

Consider I have a trait A and I have lots of structs (eg X ) impled A . Then I write many small functions(return these structs) to abstract my logic. However, the return type of these functions can be huge and too complex to write(like Chain<Map<'a, (i32, u8), u16, Enumerate<Filter<'a, u8, vec::MoveItems<u8>>>>, SkipWhile<'a, u16, Map<'a, &u16, u16, slice::Items<u16>>>> ), so I used impl A instead since as far as I know, impl <trait>s are still static dispatched:

trait A {...}
struct X {...}
impl A for X {...}
fn func_to_return_impl_A() -> impl A {...}

Now the problem is that recently I want to impl my structs to another trait B . Whenever I want to use values returned by my small functions as: func_to_return_impl_A(...).method_of_trait_B(...) , the compiler complained: no implementation for impl A <method of trait B>... . Moreover, I cannot simply rewrite the return types as impl A + B since trait B may also contain complex generic type or associated types which I cannot write myself:

impl B for X {...}

fn some_func() {
  func_to_return_impl_A().method_of_trait_B(...) //error: no implementation for impl A <method of trait B> ...
}

The only way to avoid this I can think of is to make another wrapper to wrap all my structs and impl it to trait B , but as the model getting more complex, this would be hard to maintain. How should I fix this problem?

Answer 1

Ok I think I've find the ad-hoc solution for my problem: additional Wrapper :

struct Wrapper<P: A> {
    p: P
}

impl<P: A> A for Wrapper<P> {...}
impl<P: A> B for Wrapper<P> {...}

fn func_to_return_impl_A() -> Wrapper<impl A> {...}

Then I can use methods of trait B :

func_to_return_impl_A().method_of_trait_B(...);