Eyre

Build Status Latest Version Rust Documentation

This library provides eyre::Report, a trait object based error handling type for easy idiomatic error handling and reporting in Rust applications.

This crate is a fork of [anyhow] by @dtolnay. By default this crate does not add any new features that anyhow doesn't already support, though it does rename a number of the APIs to try to make the intended usage more obvious. The magic of this crate is when you need to add extra context to a chain of errors beyond what you can or should insert into the error chain. For an example of a customized version of eyre check out jane-eyre.

My goal in writing this crate is to explore new ways to associate context with errors, to cleanly separate the concept of an error and context about an error, and to more clearly communicate the intended usage of this crate via changes to the API.

The main changes this crate brings to anyhow are

These changes were made in order to facilitate the usage of [tracing_error::SpanTrace] with anyhow, which is a Backtrace-like type for rendering custom defined runtime context setup via tracing spans.

toml [dependencies] eyre = "0.3"

Customization

In order to insert your own custom context type you must first implement the eyre::EyreContext trait for said type, which has two required methods and three optional methods.

Required Methods

This is essentially Default::default but more flexible, for example, the eyre::DefaultContext type provide by this crate uses this to only capture a Backtrace if the inner Error does not already have one.

```rust fn default(error: &(dyn StdError + 'static)) -> Self { let backtrace = backtraceifabsent!(error);

Self { backtrace }

} ```

When overriding the context it no longer makes sense for eyre::Report to provide the Display and Debug implementations for the user, becase we cannot predict what forms of context you will need to display along side your chain of errors. Instead we forward the implementations of Display and Debug to these methods on the inner EyreContext type.

This crate does provide a few helpers to assist in writing display implementations, specifically the Chain type, for treating an error and its sources like an iterator, and the Indented type, for indenting multi line errors consistently without using heap allocations.

Note: best practices for printing errors suggest that {} should only print the current error and none of its sources, and that the primary method of displaying an error, its sources, and its context should be handled by the Debug implementation, which is what is used to print errors that are returned from main. For examples on how to implement this please refer to the implementations of display and debug on eyre::DefaultContext

Optional Methods

This method is like a flexible version of the fn backtrace(&self) method on the Error trait. The main Report type provides versions of these methods that use type inference to get the typeID that should be used by inner trait fn to pick a member to return.

Note: The backtrace() fn on Report relies on the implementation of this function to get the backtrace from the user provided context if one exists. If you wish your type to guaruntee that it captures a backtrace for any error it wraps you must implement member_ref and provide a path to return a Backtrace type like below.

Here is how the eyre::DefaultContext type uses this to return Backtraces.

rust fn member_ref(&self, typeid: TypeId) -> Option<&dyn Any> { if typeid == TypeId::of::<Backtrace>() { self.backtrace.as_ref().map(|b| b as &dyn Any) } else { None } }

Once you've defined a custom Context type you can use it throughout your application by defining a type alias.

```rust type Report = eyre::Report;

// And optionally... type Result> = core::result::Result; ```


Details


No-std support

NOTE: tests are currently broken for no_std so I cannot guaruntee that everything works still. I'm waiting for upstream fixes to be merged rather than fixing them myself, so bear with me.

In nostd mode, the same API is almost all available and works the same way. To depend on Eyre in nostd mode, disable our default enabled "std" feature in Cargo.toml. A global allocator is required.

toml [dependencies] eyre = { version = "0.3", default-features = false }

Since the ?-based error conversions would normally rely on the std::error::Error trait which is only available through std, no_std mode will require an explicit .map_err(Report::msg) when working with a non-Eyre error type inside a function that returns Eyre's error type.


Comparison to failure

The eyre::Report type works something like failure::Error, but unlike failure ours is built around the standard library's std::error::Error trait rather than a separate trait failure::Fail. The standard library has adopted the necessary improvements for this to be possible as part of [RFC 2504].


Comparison to thiserror

Use Eyre if you don't care what error type your functions return, you just want it to be easy. This is common in application code. Use [thiserror] if you are a library that wants to design your own dedicated error type(s) so that on failures the caller gets exactly the information that you choose.


Incompatibilities with anyhow

Beyond the fact that eyre renames many of the core APIs in anyhow the addition of the type parameter makes the eyre! macro not work in certain places where anyhow! does work. In anyhow the following is valid.

rust // Works let val = get_optional_val.ok_or_else(|| anyhow!("failed to get value)).unwrap();

Where as with eyre! this will fail due to being unable to infer the type for the Context parameter. The solution to this problem, should you encounter it, is to give the compiler a hint for what type it should be resolving to, either via your return type or a type annotation.

```rust // Broken let val = getoptionalval.okorelse(|| eyre!("failed to get value)).unwrap();

// Works let val: Report = getoptionalval.okorelse(|| eyre!("failed to get value)).unwrap(); ```

License

Licensed under either of Apache License, Version 2.0 or MIT license at your option.


Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in this crate by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.