This pull request is being automatically deployed with Vercel (learn more).
To see the status of your deployment, click below or on the icon next to each commit.

🔍 Inspect: https://vercel.com/joystream/pioneer-testnet/3ofQGYHKEnJuyVwVymVxfa1qTW46
✅ Preview: Canceled

[Deployment for 7aea9f8 canceled]

1. I think its better to turn the entire format into one message, otherwise (se)deserializing this properly will be more cumbersome.
2. What is purpose of NonLeafNodeMerkleBranchByteOffset message?
3. CreatorPayoutBody::merkle_branches_of_non_leaf_nodes field ?

1. Do you mean both header and body should be the single message? Also, doesn't Protobuf provide out-of-the-box functionality to (se)deserializing different message properties?

For 2 & 3, please read the detailed answer:

If the payout vector is[c_0||p_0||m_0,...,c_7||p_7||m_7] , then the payload body according to proposed format would be:

c_0||p_0||proof_0||d_0 ... c_7||p_7||proof_7||d_7        

Now suppose we need to prove the membership of payout record for channel id c_2(shown green in the diagram), to prove that and claim a reward from runtime we need to pass the following args to the extrinsic

pub fn claim_channel_reward(
           ...
           proof: Vec<ProofElement<T>>, //  [`proof_3`, `proof_01`, `proof_4567`] (shown blue in the diagram)
           item: PullPayment<T>, // { c_2, p_2, m_2 }
       ) -> DispatchResult {

Problem

The payload body does not have all the Merkle branches required to construct the membership proof. For this specific example, proof_01 & proof_4567 are missing from the payload body.
To construct proof we need Merkle branches of both leaf and non-leaf(see diagram for differentiation) nodes,

Solution

Include the remaining Merkle branches in the payload body

Why need CreatorPayoutBody::merkle_branches_of_non_leaf_nodes?

merkle_branches_of_non_leaf_nodes is needed to construct the complete Merkle path required to prove the membership of payout entry(c_i||p_i||m_i) for channel id c_i.

Why need NonLeafNodeMerkleBranchByteOffset?

It is required to index into Merkle branch (of the non-leaf node) at a given byte offset in the payload body to construct Merkle path

Update

a merkle branch is a full sequence of hashes that make up the required proof

As you said in the discussion, another way to solve the problem is (and I assume this is what you actually meant):

to change the structure of payout record, considering the above example, the payout record for channel id c_2 in the payload body would be:

c_2||p_2||proof_3||proof_01||proof_4567||d_2

Schema Changes

The payout record for single channel would be:

  message CreatorPayout {
    required uint32 channel_id = 1;  // c_i
    required float cumulative_payout_owed = 2; // p_i
   
    enum Side {
        Left = 0;
        Right = 1;
    }

    message MerkleBranch {
        required bytes merkle_branch = 1;
        required Side side = 2;
    }

    repeated MerkleBranch merkle_branches = 3; 
    required string payout_rationale = 4; // d_i
  }

It will be the responsibility of someone who actually uploads the raw payload to provide the correct sequence of hashes that make up the membership proof. And, not the responsibility of client software to calculate the Merkle path and then fetch relevant branches/hashes/nodes that make up the required proof

The latest updates on your projects. Learn more about Vercel for Git ↗︎

superseded by #4334