Explicit constructor from Box<dyn Error + Send + Sync>

[vorner]

Hello

I've noticed there's a (closed) issue about having a From<Box<dyn Error + Send + Sync>> impl, that refers to wanting that from std (#66)

But would it make sense to have an explicit constructor in the meantime? Eg. Error::from_box_error(..)? That way one would have to write some kind of map_error(Error::from_box_error)? instead of just ?, but that's probably better than nothing.

[dtolnay]

anyhow! handles the conversion.

use anyhow::anyhow;

fn main() {
    let e: Box<dyn std::error::Error + Send + Sync> = unimplemented!();
    let e = anyhow!(e);
}

[vorner]

Reading the docs, doesn't this stringify the error? Won't it lose all the source chain, type information and such?

I'd hope it could somehow convert the existing dyn Error from 2-words trait object to the internal 1-word trait object. Or is it not possible?

[vorner]

Oh, I'm reading through the source code and it seems to be preserving it somehow. Would you mind if I send a PR making such usage clearer in the docs? Also, link from the Error type to the macro would make it little bit easier to find.

[rbtcollins]

Perhaps I'm missing something but anyhow! isn't very easy to use with ?.

fn main() -> anyhow::Result<()>{
   something_returning_boxed_error()?;
}

is a bit awkward - I mean:

fn main() -> anyhow::Result<()>{
  match something_returning_boxed_error(){
     Ok(v) => Ok(v),
     Err(e) => Err(anyhow!(e))
  }?
}

[vorner]

It could be shorter with map_err instead of explicit match, eg something like:

something_returning_boxed_error().map_err(|e| anyhow!(e))?

This would be even slightly more ergonomic if it existed as a function/constructor/method, eg:

something_returning_boxed_error().map_err(Anyhow::from_boxed)?

But as long as Rust doesn't allow implementing impl From<Box<std::error::Error + Send + Sync>> for Anyhow, this is probably the best that can be done :-(.