Closed kevinclark closed 3 weeks ago
anyhow
is not suitable for pacmog, I agree with that.
I think error_in_core will be stable in Rust v1.81, so I'll leave error handling alone for now.
It may be replaced when thiserror
becomes available in no_std.
Resolved in https://github.com/AkiyukiOkayasu/pacmog/pull/52 . it should be noted that Rust 1.81 or later is required. Not yet published on crate.io (there is still some work to do).
As great as
anyhow
is, it requires a global allocator. Since pacmog is targeted towards embedded systems, requiring alloc limits its adoption. Doing that for an error framework seems like an awkward tradeoff. This PR replacesanyhow
with embeded_error_chain which is no-std and no-alloc.Totally understand this is an unsolicited and somewhat large change. I need it for my use, so I'll maintain it on my own if I need to. Seemed like a reasonable speculative change though given pacmog's positioning.