Derive `Serialize` and `Deserialize` for `Error`?

[hirrolot]

Use Case:

We use sled::Error inside our own error enumeration that must implement ResponseError from actix_web. Inside impl ResponseError, we call bincode::serialize on the error, which requires it to be Serialize. Since sled::Error doesn't implement Serialize, we have to store it simply as a string.

Proposed Change:

Introduce a new feature serde and derive Serializable and Deserializable for sled::Error.

Who Benefits From The Change(s)?

sled users who want to use sled::Error in a (de)serializable context, e.g., in a microservice architecture.

Alternative Approaches

[dariusc93]

Couldnt you make a wrapper around sled::Error with your own matching enum that derives the traits you need from serde and possibly use From/Into to convert it? That make more sense to do than to request the error to support serde itself in my opinion.

[hirrolot]

I haven't tried, but possibly I could. However, it'd require quite a lot of boilerplate and would break backwards compatibility whenever a new variant to sled::Error is added. From the library side, it'd be much easier to just add #[derive(Serialize, Deserialize)].

[quietlychris]

I'm not actually sure it works this way, at least trivially using the NewType pattern

#[derive(Serialize, Deserialize)
struct SledError(sled::Error);

will still have issues. serde's API has been stable for years now as 1.0 + SemVer, and at least within the projects that I work on, it does seem pretty common to create an opt-in feature for exposing serde traits for external types like sled::Error. I just ran into this issue myself, and it would make things much easier. From a development point of view, it shouldn't be a big lift to either add or maintain as well, and I'll probably look into submitting a PR about this once the new bloodstone work that Tyler's merging gets more fully rolled out.