Currently, we just have a (slot, root) -> parent index stored in the database. We should instead keep this information in an in-memory tree for easier traversal. This also enables weight caching for head and safe target computation.
For reference, Lighthouse uses a "proto-array" implementation: https://github.com/sigp/lighthouse/tree/stable/consensus/proto_array/src
while Prysm uses a doubly-linked tree: https://github.com/OffchainLabs/prysm/tree/develop/beacon-chain/forkchoice/doubly-linked-tree
On a second version, we can add a DB fallback for long periods of non-finalization.
Related to #84
Currently, we just have a (slot, root) -> parent index stored in the database. We should instead keep this information in an in-memory tree for easier traversal. This also enables weight caching for head and safe target computation.
For reference, Lighthouse uses a "proto-array" implementation: https://github.com/sigp/lighthouse/tree/stable/consensus/proto_array/src
while Prysm uses a doubly-linked tree: https://github.com/OffchainLabs/prysm/tree/develop/beacon-chain/forkchoice/doubly-linked-tree
On a second version, we can add a DB fallback for long periods of non-finalization.
Related to #84