What you link to talks over the great `unsafe` drama, but that's not cheating. `unsafe` is in Rust for a good reason. Every C and C++ codebase is effectively one big `unsafe` block from the perspective of Rust developer, and we don't think C/C++ are cheating. Rust devs frown upon overusing `usafe` for a good reason: it leads to obscure bugs and security issues. But if you use it correctly, in place that is critical for your performance - go for it.

Having said that - a lot of frameworks in that benchmark are over-optimizing in ways that no real practical app would do and arguably are cheating.If there's anything in particular about actix here, please let me know - I'm honestly interested in it.

Note that this is for the "raw" version of the plain-text and JSON benchmarks, not the fortune benchmark covered in the OP. But I wonder why that raw version is even there. It certainly isn't "realistic", as it claims to be in benchmark_config.json

> In fairness, it seems other benchmarks are doing these things

Personally, I think it's pretty reasonable to sink to the same level as the competition.