How does it compare to nuclei

[NoelJacob]

How does nuclei compare to this? To be frank, I don't understand nuclei or this that good. Any benchmarks possible?

[NoelJacob]

Also another question, Monoio says this on their README.md:

Monoio can not solve all problems. If the workload is very unbalanced, it may cause performance degradation than Tokio since CPU cores may not be fully utilized.

Am I right in guessing this is also valid for compio?

[George-Miao]

How does nuclei compare to this? To be frank, I don't understand nuclei or this that good. Any benchmarks possible?

Its io-uring impl is unsound, and I don't know about other parts. Its API does not count into completion-based IO (so does entire Rust eco system, basically) and does nothing to guarantee the safety (e.g. by adding a shim buffer like we do), so...

[Berrysoft]

Another notable issue is that, although it claims to support Windows, the code is actually in a PR, still not merged.

[Berrysoft]

Am I right in guessing this is also valid for compio?

Well, yes. Compio is not a silver bullet.

CPU balance is always a hard problem. On Linux, io-uring is optimized for single-threaded runtime, which means the CPU is unbalanced if the workload is unbalanced.

In addition, if you wouldn't like the runtime taking the ownership of the buffers, tokio is still the better choice, because the language design of Rust is not fully suitable for completion-based async.

[NoelJacob]

Can we use it with rayon or can compio be used like a rayon alternative?

[Berrysoft]

Rayon is for multi-threaded computing while compio is for async IO. They are not for the same purpose.

You can always use compio with rayon. Just start a new thread or use spawn_blocking and use rayon in the new thread. It's not a good idea to use rayon inside async context.