Inference regression about error-chain

[Yamakaky]

Yeah, I know. Since ChainErr was replaces by the crate-level https://docs.rs/error-chain/0.6.1/error_chain/trait.ResultExt.html, chain_err doesn't know which ChainedError it should return. Not sure what do to to fix that without re-adding ChainErr.
As a temporary fix, you can do

let res: Result<_> = result.chain_err(..);
res.unwrap();

[jsen-]

Unfortunately, the same occurs for example here:

#[macro_use]
extern crate error_chain;
use error_chain::ResultExt;
error_chain! { }

fn real_main() -> Result<()> {
    ::std::fs::create_dir_all("path/to/some/dir")
        .chain_err(|| "Unable to create some directory")?;
    // do some more stuff
    Ok(())
}

fn main() {
    real_main().unwrap();
}

My code is littered with such constructs 😟

[Yamakaky]

Similar issue here.

[brson]

This rings a bell and may have been part of the original motivation for having the trait be per crate.