feat: make sure the error kind is properly transferred when serializing io errors

[rklaehn]

Description

I now go through a lot of effort assigning good error kinds and don't want them to be lost!

Breaking Changes

None, except the rpc API will behave slightly differently. But we don't care for that level of stability yet!

Notes & open questions

Note: you can cast an ErrorKind to u32, but the order in the enum is not stable. Hence the boilerplatey match.

Fixes #110

Change checklist

• Self-review.
• Documentation updates following the style guide, if relevant.
• Tests if relevant.
• All breaking changes documented.

[github-actions]

Documentation for this PR has been generated and is available at: https://n0-computer.github.io/iroh-blobs/pr/111/docs/iroh_blobs/

Last updated: 2025-07-11T09:33:12Z

[matheus23]

Eh, I wish there were a way to check for future variants that might get added and have a test failing or sth like that. But with #[non_exhaustive] and no way of listing all current types of errors, I wouldn't know how.

(perhaps some hackery with mem::transmute between u32 and ErrorKind and counting up until it fails? 😅 )

[rklaehn]

Eh, I wish there were a way to check for future variants that might get added and have a test failing or sth like that. But with #[non_exhaustive] and no way of listing all current types of errors, I wouldn't know how.

(perhaps some hackery with mem::transmute between u32 and ErrorKind and counting up until it fails? 😅 )

Yeah, I actually tried to leave out the _ case, but the compiler shouted at me due to non_exhaustive...