trie: missing trie node types

[qdm12]

Task summary

https://github.com/paritytech/substrate/blob/ded44948e2d5a398abcb4e342b0513cb690961bb/primitives/trie/src/node_header.rs#L85-L107

we only implement leaf, branch without value and branch with value types, so it appears some types are missing.

Source comment

[noot]

which types are missing? HashedValueLeaf and HashedValueBranch? also do we actually need these types? the gossamer trie doesn't need to be the exact same as the substrate one

[seunlanlege]

the gossamer trie doesn't need to be the exact same as the substrate one

any reason for this?

[danforbes]

@noot - are you able to provide an answer to Seun's question above?

[noot]

@seunlanlege @danforbes what I meant was that the implementation doesn't need to match in terms of exact types and the way it's implemented, as long as both implementations match the specification and result in the same merkle value for the same input. in general I think the less we reference substrate the better, and we definitely shouldn't blindly copy types or functions from substrate without understanding what they do.

[danforbes]

That makes sense. I agree that we don't need to use the same naming conventions and internal structures as Substrate; I'm unclear as to whether or not Gossamer's trie implementation is missing certain types of nodes, etc that inform part of the trie's actual specification.

[noot]

turns out this is related to #2392 - the new types are for state trie version 1, not the state trie version 0 (what's currently implemented)

[noot]

closing in favour of #2418