Ouput and input commitments are reversed in mod.rs about grin HOT 3 CLOSED

As long as it's consistent it doesn't really matter. Signs make sense in algebra but aren't equivalent when translated to elliptic curve points. In the code I chose to follow the Jedusor paper, where one takes the r-value of change minus inputs and outputs are added afterward. Makes sense?

from grin.

I'm not quite sure I understand. In the code, the signature of commit_sum is:
pub fn commit_sum(&sef, positive: Vec<Commitment>, negative: Vec<Commitment>) -> Result<Commitment, Error>
So commit_sum(inputs, outputs) is going to do inputs - outputs of the transaction we want to verify, and not outputs - inputs, which will give the inverse (x, -y) if the key kG is (x, y). You can try inverting the arguments, the y of the public key changes depending on their order. Sorry if I'm missing something here, but the excess is defined as outputs - inputs and not inputs - outputs right ?

from grin.

You're right, for some reason I read that line secp.commit_sum(output_commits, input_commits) instead of what it actually was (secp.commit_sum(output_commits, input_commits)). I must have been looking at it too late.

I'm about to push a fix to invert everything. As I said it doesn't really matter as far as validity is concerned but it will be more consistent with the literature.

from grin.