Closed danielmasny closed 3 months ago
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 91.68%. Comparing base (
2f873e1
) to head (99841d0
).
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
I disagree with this design. I do not think compute proof should be responsible for counting the elements added to it. Look at all the complex tuple return types and the frequently ignored values. This is not a good design.
I think when we perform a multiplication and push it into the list of multiplications that need to be verified, THAT is the place to increment a counter. In fact, it's only that first stage where we need to perform this count.
@benjaminsavage I would be fine with that too and was thinking about that as well. A reason why I think this PR is more straightforward is that we just count the amount when we actually compute the proof (which will be the same for all three proofs and doesn't need to be sent over the network). Counting the multiplications when they are stored comes with two challenges:
This PR allows to keep track of amount of u,v values when generate proof. This is important since the verification will need to know the amount.