Proof Trait Todos

[spaceships]

This issue tracks our progress on the following tasks:

• Comment code: why we used From<Statement> in mithril_proof
• Comment: why Rc is used in mithril_proof::Statement
• Code: change the output of Witness::verify to use a Result type so we can know more about what failed than false.

[abakst]

@markulf you had a question about why we used From, noting that the Proof Statement and Witness types will always be the same as the Mithril-specific statement.

I'm not sure if I understand: while this is true for the trivial proof system, if a generic BulletProof system (assuming it exists) implements Proof, then I would imagine its Statement and Witness types would be something like vectors of circuit inputs, which differs quite a bit from the Mithril-specific type.

Does that make sense?

[abakst]

@iquerejeta @markulf we do not necessarily need Rc for all of the Statement fields -- it's actually most useful for (1) avoiding parameterizing Statement by a lifetime and (1) avoiding the need to copy the whole MerkleTree (it's probably fine to copy/clone the remaining fields). If Statement takes a lifetime parameter, then the lifetimes of the various proofs end up being tied to the lifetime of the StmClerk, which isn't so pleasant.

We'll certainly include this explanation in the documentation. I think we will also want to remove Rc from the other fields.

[markulf]

@markulf you had a question about why we used From, noting that the Proof Statement and Witness types will always be the same as the Mithril-specific statement.

I'm not sure if I understand: while this is true for the trivial proof system, if a generic BulletProof system (assuming it exists) implements Proof, then I would imagine its Statement and Witness types would be something like vectors of circuit inputs, which differs quite a bit from the Mithril-specific type.

Does that make sense?

Yes, makes sense. Can we clarify in comments which would be the high-level statement and witness and which the constraint system specific low-level one and who does the mapping.

[abakst]

I have opened #14 to address some of the points in this issue, but will open another PR to implement richer return types.