mssmt: Add CompressedProof serialization #29
Conversation
wpaulino
force-pushed
the
compressed-proof-encoding
branch
2 times, most recently
from
02e2e4d
to
9899774
Compare
This is currently unused, but will serve useful when introducing proof serialization. Proof nodes have their NodeKey and NodeSum serialized without their pre-image, so we need a way to reconstruct a Node without them.
wpaulino
force-pushed
the
compressed-proof-encoding
branch
from
9899774
to
4299194
Compare
We don't add one for the decompressed variant as there's no reason to
serialize those when the compressed variant can yield a more efficient
serialization.
The serialization scheme for CompressedProof is as follows:
* Nodes:
* Big endian encoding of the number of nodes present as uint16
* Each node:
* 32 byte key
* Big endian encoding of the node sum as uint64
* Bits:
* 32 bytes of the compressed bits (256 bits total)
The SplitCommitmentRoot struct only existed as we wouldn't be able to recover it without knowledge of the node's pre-image. Now that mssmt.ComputeNode exists, this is no longer an issue.
wpaulino
force-pushed
the
compressed-proof-encoding
branch
from
4299194
to
6d33f92
Compare
| @@ -2,10 +2,6 @@ package mssmt | |||
|
|
|||
| // Proof represents a merkle proof for a MS-SMT. | |||
| type Proof struct { | |||
| // Leaf is the leaf node the proof is valid for. If the leaf is empty, | |||
There was a problem hiding this comment.
What's the motivation here? Making it more similar to Bitcoin's control block structure? (where the leaf is actually obtained from elsewhere in the stack)
There was a problem hiding this comment.
Yeah and also noticed that in most cases we already have the raw leaf as a serialized Asset so it wasn't really being used anyway.
There was a problem hiding this comment.
Yeah makes sense, you typically already know the leaf that you want to check inclusion of, or that can be transmitted prepended to the proof itself.
| // ComputedNode is a node within a MS-SMT that has already had its NodeKey and | ||
| // NodeSum computed, i.e., its preimage is not available. | ||
| type ComputedNode struct { | ||
| key NodeKey |
There was a problem hiding this comment.
Ultimately I think the node key terminology here is pretty confusing. It's actually the node digest/hash, but it reads as though it's somehow the key used to traverse the tree to arrive at the given node.
There was a problem hiding this comment.
Agreed, I would be ok with s/NodeKey/NodeDigest/g before we merge any other in-flight PRs.
There was a problem hiding this comment.
Don't think this is blocking though, will likely do another sweep afterwards to clean up naming.
| @@ -181,7 +181,7 @@ func (t Tree) MerkleProof(key [hashSize]byte) *Proof { | |||
| // VerifyMerkleProof determines whether a merkle proof for the leaf found at the | |||
| // given key is valid. | |||
| func VerifyMerkleProof(key [hashSize]byte, leaf *LeafNode, proof *Proof, | |||
| root *BranchNode) bool { | |||
| root Node) bool { | |||
There was a problem hiding this comment.
What do we gain by making this the interface? If it's a leaf, then this operation can never succeed.
There was a problem hiding this comment.
This would allow us to use asset.SplitCommitmentRoot (which is now of type mssmt.ComputedNode directly without doing a type conversion.
|
add multi address send RPC and itest
We don't add one for the decompressed variant as there's no reason to serialize those when the compressed variant can yield a more efficient serialization.
The serialization scheme for CompressedProof is as follows: