Asuna

### Before ``` test bench_1_file ... bench: 195.69 ns/iter (+/- 13.53) test bench_2_file_async ... bench: 243.52 ns/iter (+/- 15.84) test bench_3_rotating_file_size ... bench: 197.99 ns/iter (+/- 35.30) test bench_4_rotating_daily ......

Okay, slightly improvement, since this doesn't affect any public API, so we are not in a hurry to include it in v0.5.0.

Turns out `itoa` doesn't support leading zeros at the moment, I have opened a issue for it. dtolnay/itoa#61 Since we are using formats like `{:03}`, it looks like `numtoa` is...

The `{integer}_padded` functions in `numtoa` only supports padding length >= `{integer}::BUF_LEN`, it's still hard to replace std formatting like `{:03}` for integer type `u32` with `numtoa`. I have opened a...

> What do you mean „safely“? I mean soundness. --- > Is it actually worth it? Is there a good enough use case for such a thing? I was trying...

I took a look just now, there are some issues: - To get an immutable (const) pointer from a `Pin`, either we bound `T: Unpin` then call `Pin::into_inner`, or we...

Or perhaps I'm overthinking it, getting pointer from `Pin` is safe -- does not violate pinning invariants, as seen examples in stdlib. https://github.com/rust-lang/rust/blob/d427ddfe90367eaa6d2ed7bb8a16559f0230f47a/library/std/src/sys/sync/once_box.rs#L63 https://github.com/rust-lang/rust/blob/d427ddfe90367eaa6d2ed7bb8a16559f0230f47a/library/std/src/thread/thread.rs#L279 And converting `*const T` to `*mut...

I have the same issue, here is my code (I apologize for not having time to minimize it at the moment): ```rust use std::{env, path::PathBuf}; use bindgen::callbacks::{ AttributeInfo, DeriveInfo, EnumVariantValue,...