[feat] Utreexo: adding new item to the accumulator #167

[Jeanmichel7]

• Resolves [feat] Utreexo: adding new item to the accumulator  #167
• follows contribution guide
• code change includes tests

[vercel]

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

Name	Status	Preview	Comments	Updated (UTC)
raito	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Sep 18, 2024 1:57pm

[Jeanmichel7]

ok because we have to remove it from the cache when we delete an utxo

[m-kus]

@Jeanmichel7 cache is a separate task #164 you don't need to implement that

[Jeanmichel7]

@m-kus @maciejka @TAdev0 I wonder if the type we want to add to the utreexo accumulator isn't just the hash of the output (without the cached field)?
It's when you go through the array of outputs of a transaction, that you add the leaves to the available utxo set.
So we don't need to provide an OutPoint but a TxOut to the add function?

The OutPoint is only useful for validate and delete the utxos available in the accumulator provided in a transaction's input.

Did I understand correctly?

I didn't know how to test the logic of the accumulator, so I tried to use a dict which is very close to the basic algo to compare the diff.

[m-kus]

We need the entire outpoint because we rely on extra fields like block height/time and iscoinbase for finality checks.

If this data is not in a leaf or cache it is unconstrained, ie you can provide arbitrary values and the program won’t be able to tell

[Jeanmichel7]

So we assume that when we add an OutPoint, the cache in data is always set to true? otherwise it could create 2hash for the same UTXO?

[m-kus]

It’s a bit more complex, but we’ll get there gradually. Generally we want to ensure two things:

• All referenced outpoints are constrained (either via utreexo or cache)
• all the referenced outpoints are spent exactly once

When we handle a transaction input there might be two cases:

• it is cached, we need to check that it actually exists in the cache and then we remove it
• otherwise we check that it hasn’t been spent yet, and mark for removal

For every output we construct the outpoint and either put it in the cache or mark for addition.

In the end we do batch inclusion verification/removal, the add all the new leaves.

[TAdev0]

resolves #167

doesnt work when its not in the first comment

@Jeanmichel7 can you add it on the first comment of the issue? so that it links your PR to the issue , and it will also automatically close the issue once merged

[m-kus]

LGTM
Some refactoring is needed (e.g. it is clear now that we can separate UtxoSet and Utreexo accumulator), but it's better to do after we merge all pending PRs