About

This is a program that runs benchmarks of common basic operations of using SNAFU vs Anyhow (both of which are error-helping libraries for Rust), to get a sense of how their speeds compare for these operations in isolation, where the purpose of Anyhow overlaps with that of SNAFU (SNAFU has a broader purpose but aims to support similar use-cases as Anyhow), and this is focused on the types snafu::Whatever vs anyhow::Error (actually, this uses my tweak of Whatever to make backtrace capturing optional like it is for anyhow::Error).

How to run

cargo run --release -- --bench

Or:

FILTER='Vs|(Thin|Fat)-More|Empty'  # Or whatever regex you want.
cargo run --release -- --bench "$FILTER"

Or:

RUST_LIB_BACKTRACE=1  cargo run --release -- --bench ...

Etc.

Note

In contrast to these benchmarks, in some other real-world situations, additional computations would be done around when error-handling operations are, and so the comparative speed of how these libraries do the same/similar operation is sometimes only a very small fraction of the total cost of everything that's being done, and so the impact of one library's approach being somewhat slower is sometimes negligible for such situations.

TODO

I'm not super familiar with these libraries nor with using them, nor with Criterion.rs, and I might have missed how to setup the most-appropriate comparisons of most-similar operations as well as they should be. If so, I'd like to improve this. Feel free to submit PRs.