Using `array-bytes` for Array/Bytes/Hex Operation

[aurexav]

Hi! I'm the author of array-bytes.

This crate is completely optimized for Substrate development.

It's pretty useful and easy to use. And it's completely no-std.

Did you consider replacing hex crate, HexDisplay struct, and any other array/bytes/hex operation with this crate?

[bkchr]

This crate is completely optimized for Substrate development.

Can you elaborate?

[aurexav]

Can you elaborate?

Sorry, let me elaborate on this.

There is no performance optimization. But this crate makes the developer's life much easier (IMO).

It is completely no-std. So, you don't need to write something like array-bytes/std in the toml file.

It provides a lot of handy functions.

For example, the sp_core::H256 has implemented the From<[u8; 32]>. With this crate, you could write something like:

// better error handling
let hash = array_bytes::hex_into::<H256, _>("0x840cff528f0a427cf1405cc474c76b369890fc01951c73c71ae1e2ad1e8baa7d").expect("valid hex; qed");
// also provides an infallible method
let hash = array_bytes::hex_into_unchecked::<H256, _>("0x840cff528f0a427cf1405cc474c76b369890fc01951c73c71ae1e2ad1e8baa7d");

// For `AccountId32`
let account = array_bytes::hex_into::<AccountId32, _>("0x840cff528f0a427cf1405cc474c76b369890fc01951c73c71ae1e2ad1e8baa7d").expect("valid hex; qed");
let account = array_bytes::slice_into::<AccountId32, _>(&[100, 118, 109, 58, 0, 0, 0, 0, 0, 0, 0, 213, 244, 148, 7, 4, 235, 76, 229, 228, 181, 24, 119, 212, 155, 88, 163, 233, 53, 49, 182, 96]).expect("valid data; qed");

And it supports a dynamic hex2bytes method. Not like hex crate, it requires a &'static str at compile time.

And in the runtime development, we work with the rust array/slice/vec every day. It provides some functions to transform the dynamic type into a fixed length array, example.

Furthermore, if you are working with offchain-worker or genesis data. You might need to interact with the JSON object. It provides some useful deserializing functions. For example:

"header": {
    "block_number": "0x123"
}

#[derive(Debug, PartialEq, Deserialize)]
struct Header {
	#[serde(deserialize_with = "array_bytes::de_hex2num")]
	block_number: u32,
}

So, I hope Substrate could use this crate. And let more Substrate developers know/enjoy such convenient functions/tools.

[bkchr]

I'm fine with switching to this crate.

[gilescope]

https://github.com/hack-ink/array-bytes/blob/38fb70e60306f2d6096bc2453fbcec150c59237c/src/lib.rs#L243 - Shouldn't these unchecked functions be marked as unsafe functions? I should not be able to call them from safe code because they are not safe.

[aurexav]

hack-ink/array-bytes@38fb70e/src/lib.rs#L243 - Shouldn't these unchecked functions be marked as unsafe functions? I should not be able to call them from safe code because they are not safe.

In https://paritytech.github.io/substrate/master/sc_service/index.html?search=unchecked, there are also a lot of unchecked functions.

IMHO, unchecked is a marker, you should know what you are doing.
For me, I only use unchecked functions in tests or some panic-ble parts.
Just like you can write .unwrap() in runtime code but you do not.

But any discussion is welcome.

[bkchr]

This function is not really unchecked as @gilescope is saying. You are using there transmute without checking the input. You should use str::from_utf8.

[aurexav]

This function is not really unchecked as @gilescope is saying. You are using there transmute without checking the input. You should use str::from_utf8.

For this context (hex conversion), str::from_utf8 do a lot of unnecessary checks. I'll consider marking this function as unsafe.

[aurexav]

Thanks for the code review.

• Optimize instruction size hack-ink/array-bytes#14
• Mark hex_bytes2hex_str_unchecked as unsafe hack-ink/array-bytes#15

[shawntabrizi]

Would make sense to me to have SR Labs or someone else audit the crate before we switch to it?

[aurexav]

Would make sense to me to have SR Labs or someone else audit the crate before we switch to it?

I'm looking forward to this.