Open jorgeantonio21 opened 1 year ago
yes to this issue;
but on update
and delete
:
update and delete operations are not CALM https://arxiv.org/abs/1901.01930.
So I would leave implementing them for when we have enough structure on the commitments of (inserted / appended) records, bootstrapped a peer-2-peer network, and possibly even the full counterfactual proofs, because all these components may weigh in on how we want to think about not-CALM operations (and where our optimization focus will lie)
I would put these ops on ice until much more has been built out
I am not sure we should multithread the connection to the anchorDB; rather improving the runtime; and multithreading different tasks first might be more important first; we will want to handle multiple concurrent packages soon, then sharing the anchorDB connection will be important; putting this on ❄️ ice for now
Atm, we only deal with insert operations on the DB. We might as well write the flush logic to deal with multiple threads, to improve efficiency of runtime execution. @vadim-frolov suggested we use connection pools for it:
https://docs.rs/r2d2/latest/r2d2/index.html
Further discussion might be needed for update and delete operations.