Seems n² but could be n.log(n)

[Ten0]

Unless I'm mistaken the implementation of first fit is n² because it checks every bin for every item,

pack_it_up/src/online/first_fit.rs

Lines 27 to 39 in 3b74472

	for item in items.into_iter() {
	// Find the first bin that the item fits in
	match bins
	.iter_mut()
	.filter(|bin| item.size() <= bin.remaining_capacity)
	.next()
	{
	Some(bin) => bin.add(item),
	None => bins.push(Bin::with_item(bin_size, item)),
	}
	// TODO: Should be move bins that are full to a new vector to avoid having to iterate them?
	}

And this could be improved to n.log(n) (complexity could be improved but for small sizes this probably isn't better) by storing the current size of each bin in a BTreeSet<(usize, usize)> where tuple is (size_of_the_bin, index_of_the_bin).
Using the BTreeSet::range function enables you to look up the best fit in O(log(n)) in the map.