modulitos / bin_packer_3d Goto Github PK

Some possible options:

• Github Actions
• Circle CI
• Travis
• Gitlab CI

Possible resource:
https://www.lpalmieri.com/posts/2020-06-06-zero-to-production-1-setup-toolchain-ides-ci/#5-continuous-integration

We wanted to support floats as dimensions, but I couldn't figure out an easy way to do without aliasing our Dimension type to f32: pub type Dimension = f32;. Doing that forces us to always use floats as argument constructors, even in the case where an int is preferable:

    let deck = Item::new("deck", [2.0, 8.0, 12.0]);
    let die = Item::new("die", [8.0, 8.0, 8.0]);
    let items = vec![deck.clone(), deck.clone(), die, deck.clone(), deck];

    let packed_items = packing_algorithm(Bin::new([8.0, 8.0, 12.0]), &items);

I think the desired use case is to have the ability to support ints as well as floats:

    let item_1 = Item::new("deck", [2, 8, 12]);
    let item_2 = Item::new("deck", [2.25, 8.34, 12.65]);

But I'm not sure about the best way to do this. Perhaps we should leverage this crate: https://docs.rs/rust_decimal/1.0.1/rust_decimal/ ? Or maybe roll our own implementation?

If using our own implementation, one idea is to create a new struct with a fixed precision of 2 decimal places:

struct Dimension {
  whole_number: u8, // 0-255
  decimal: u8, // max value of 99
}

This would give our dimensions a range of 0.00 to 255.99, and should be easy extend this further as needed. However, in order to support the current new function:

https://github.com/modulitos/bin_packer_3d/blob/master/src/block.rs#L28-L32

pub fn new(d1: Dimension, d2: Dimension, d3: Dimension) -> Self {

we'll need to convert the integers and floats into our new Dimension struct. Perhaps we can have Dimension implement the From (or Into) traits:

impl From<u8> for Dimension {...}
impl From<f32> for Dimension {...}

But I'm not sure if this is the right approach - does this seem reasonable? Any feedback here would be appreciated 🙏