binrw_lists.rs

use binrw::binrw;
use binrw::io::{Cursor, Seek, SeekFrom};
use binrw::{BinReaderExt, BinWriterExt};

#[binrw]
#[derive(Debug)]
struct Vec3 {
    x: f32,
    y: f32,
    z: f32,
}

#[binrw]
#[derive(Debug)]
struct Mesh {
    #[bw(calc = (vertices.len() - 1) as u32)]
    vertex_count: u32,

    #[br(count = vertex_count + 1)]
    vertices: Vec<Vec3>,
}

fn main() {
    let mesh = Mesh {
        vertices: vec![
            Vec3 {
                x: 1.0,
                y: 2.0,
                z: 3.0,
            },
            Vec3 {
                x: 11.0,
                y: 12.0,
                z: 13.0,
            },
            Vec3 {
                x: 21.0,
                y: 22.0,
                z: 23.0,
            },
        ],
    };

    let mut writer = Cursor::new(Vec::new());
    writer.write_le(&mesh).unwrap();

    let mut reader = writer;
    reader.seek(SeekFrom::Start(0)).unwrap();

    let _: Mesh = dbg!(reader.read_le().unwrap());
}

why it need -1 and +1 ?

@rain2307 merely a demonstration that you can do arbitrary computation in those positions. It effectively translates to "the length as written to disk is one less than the actual length". Not something a normal format would do, but arbitrary computation is useful for implementing arbitrarily complex file formats