Structs and enums in Rust

Whenever a Rust circuit is compiled, rustc applies various optimizations to reduce its memory usage.

Among these memory optimizations is reordering fields in structs and enums to avoid unnecessary 'paddings' in circuit IRs. Consider the following example:

use ark_pallas::Fq;

type BlockType = [Fq; 2];

#[derive(Copy, Clone)]
pub struct BlockDataType {
    prev_block_hash: BlockType,
    data: BlockType,
    validators_signature: i32,
    validators_key: u64,
}

The public input representation of the BlockDataType struct would look as follows:

"struct": [
    {
        "array": [{"field": 1}, {"field": 1}]
    },
    {
        "array": [{"field": 3}, {"field": 1}]
    },
    {
        "int": 1
    },
    {
        "int": 1
    }
]

When compiling the BlockDataType struct, rustc may reorder its fields.

When assigner is called on a circuit with this struct, the circuit IR would conflict with the public input as the field order in the IR and the public input file would no longer match.

To avoid this problem, use the #[repr(C)] directive:

use ark_pallas::Fq;

type BlockType = [Fq; 2];

#[repr(C)]
#[derive(Copy, Clone)]
pub struct BlockDataType {
    prev_block_hash: BlockType,
    data: BlockType,
    validators_signature: i32,
    validators_key: u64,
}

If this directive is included, Rust will treat structs and enums as C-like types, meaning that rustc will never reorder fields in them.