Still learning! Anyone know whether there's a better way to avoid this longwinded series of T:from_i32 statements for an algorithm that works with both variables and constants?

thread 'main' panicked at 'assertion failed: `(left == right)` (left: `10900.115611845817`, right: `10900.115612674515`)', <anon>:8

Something's going to have to be done about that.

Please remove the warnings.

Indeed, I'm very aware of the error, and will work on fixing it, nobody is expecting this PR to be merged with failing tests :)

Anyone know the answer to the T::from_i32 question, though?

I've had the same issue when using "actual numbers" within generic functions, and haven't been able to get around it, but I also haven't taken the time to ask anyone who works on/with num. It's probably worth asking on #rust, or users.rust-lang.org.

OK, so the consensus seems to be something like:

fn f<T: FromPrimitive>(i: i32) -> T { T::from_i32(i).unwrap() }

I'd assumed it would be more complicated than that.

Just rebased this off master. @tmcw It looks like there are failing tests. Do you know why they're failing? It'd be great if this could merge before #85, but no worries if you're short on time, just let me know :)

@tmcw @frewsxcv this is most likely related to comparing floats using assert_eq(). we should probably use within_epsilon() for the assertions (see https://github.com/georust/rust-geo/blob/master/src/algorithm/distance.rs#L347)