Most of the time, you get blocks of note commitments that do not contain new received notes. In that case, you just want to update the witnesses with the new block data. Normally, that would require hashing the new note commitments and updating the commitment tree.
Bridges are pre-calculated hashes that enable us to update the witnesses without computing any hash.
Using a Bridge is much faster than doing the computation normally.
Even better, the bridges are calculated server side and can be used by any client anonymously, i.e. they don't leak privacy.
With a Bridge, we want to capture the data needed to update witnesses and to update the tree state. We are not required to handle new received notes therefore we do not need the internal hash values.
- For the witnesses, we need the first node of each level if it is the right node of a pair1.
- For the tree state, we need the last node of each level if it is the left node of a pair2.
Merging Bridges
Continuous bridges can be merged by taking the head of the first bridge and the tail of the second one.
Applying a Bridge
After a bridge is applied to the witnesses and the tree state is updated, we can continue processing nodes normally.
Warp 2 mixes bridges and normal processing. We receive compact blocks that have both data, we search new received notes.
If we find any, we use the normal processing path since we have to add new witnesses. Otherwise, we use the bridge to quickly update the existing witnesses.
At the end, the tree reaches the same state.