Integrate the trie

[Avi-D-coder]

This PR just sets up the trie thread and state.
We still need all the prior logic against the btree to pre-validate transactions.
That btree will be replaced by a cache db (PublicKey -> Account).

We could remove pre-validated transaction requirement with an immutable trie::SnapshotBuilder that supported could rollback.
However, we would still need to prevent DoS with the cache db. Real hardware is very different from the zkVM, binary tries don't perform all that well on SSDs.

In the future we will also need to warm up the NodeHash -> TrieNode DB right before a txn hits the trie state thread. By cashing TrieNodes into ram we can move slow IO off that thread.

Next Steps after this PR:

• Adding persistence with rocks
• Adding a validating thread for testing against (like a mock contract for testing)
• General testing txns (we can replicate what I did in the trie repo).

[Avi-D-coder]

I'm going to merge the txn logic between the trie and btree-cache-db before this should be merged.

[github-actions]

File	Coverage
All files	54%	❌
kairos-crypto/src/implementations/casper.rs	3%	❌
kairos-test-utils/src/cctl.rs	87%	✅
kairos-test-utils/src/kairos.rs	95%	✅
kairos-server/tests/transactions.rs	85%	✅
kairos-server/src/routes/deposit.rs	88%	✅
kairos-server/src/routes/transfer.rs	90%	✅
kairos-tx/src/asn.rs	76%	✅
kairos-tx/src/error.rs	0%	❌
kairos-server/src/config.rs	0%	❌
kairos-server/src/errors.rs	12%	❌
kairos-server/src/state.rs	90%	✅
kairos-server/src/utils.rs	22%	❌
kairos-test-utils/src/cctl/parsers.rs	66%	✅
kairos-server/src/state/transactions.rs	57%	❌
kairos-server/src/state/trie.rs	35%	❌
kairos-server/src/state/transactions/batch_state.rs	42%	❌

Minimum allowed coverage is 60%

Generated by 🐒 cobertura-action against 4635140

[marijanp]

Minor things, mostly requests for additional documentation

[marijanp]

Could you document what new does

[koxu1996]

Very solid PR 👏. One requested change is to remove epoch locking - as we discussed during last deep dive - so we avoid accepted transactions that are unexpectedly failing because of batch end/limit. 💬 (This was already done, I missed it somehow.)

Also thanks for the improvement to kairos-tx, I will extract them to separate PR to avoid clutter in yours.

[Avi-D-coder]

One requested change is to remove epoch locking - as we discussed during last deep dive - so we avoid accepted transactions that are unexpectedly failing because of batch end/limit.

I think I already did this in 74d0444.

I removed the my outdated comment about batch epoch.
Although I think we might still need an epoch for time expiring transactions.

[koxu1996]

What about TTL? It could be maximum timestamp, or maximum epoch.

[marijanp]

Where I asked for more docs for types and functions, I was more in the hope of getting explanations of how the different datastructures are supposed to be used together, what caveats there are.

[marijanp]

All of this work in this file just to add a nonce to some of the types. I think we need to think how we can refactor kairos-tx

[marijanp]

Maybe get rid of SigningPayload and adding the nonce where relevant.

[Avi-D-coder]

I'd like to move nonce to transfer and withdraw in the future.

[Avi-D-coder]

However some of this duplication is to make type inference nicer.

SigningPayload::new vs SigningPayload::new_*

[Avi-D-coder]

Where I asked for more docs for types and functions, I was more in the hope of getting explanations of how the different datastructures are supposed to be used together, what caveats there are.

I added a bit more, but I'm not clear on what parts are confusing?
Are you looking for an overview of the trie logic or something else?

[koxu1996]

🚀

[marijanp]

Thank you for adding the comments