CCF
CCF copied to clipboard
microsoft
Confidential Consortium Framework
`Store::commit` currently does something like: ``` commit(new_tx) { pending_txs.queue(new_tx); batch = build_contiguous_prefix(pending_txs); if (!batch.empty()) { consensus.replicate(batch); } } ``` Each call is really doing 2 things: extending a queue, and...
The underlying implementation of the a key-value map currently uses a CHAMP, which is fast compared to the original Red-Black tree implementation (see benchmark results: https://github.com/microsoft/CCF/issues/2814#issuecomment-940099829), but does not provide...
We currently make extensive use of SHA256, which is hard-coded in many parts of the codebase. We should make this user-configurable and offer e.g. SHA512 and SHA3.
Follow up from #1648. Now that we have a mechanism to recover historical ledger secrets from the ledger, we shouldn't keep in enclave memory all the historical ledger secrets that...
Follow up from #1988, but with all strings created with `fmt::format()`, e.g. error messages.
We currently have an unnecessarily strict connection between the prefix of a URL path and the auth needed to access that path. A CCF service is divided into 3 frontends...
In a three node network: 1. Isolate primary 2. Issue transactions against isolated primary that will not commit, wait until they get chunked to disk 3. Retire former primary 4....
Unearthed by https://github.com/microsoft/CCF/issues/3752 When the private ledger is being recovered, the service should also be resilient to elections. While we expect this to work in some cases, I believe this...
Follow-up to #3491/#3497. There are a few other parts of the code that look very suspicious to me. For instance: https://github.com/microsoft/CCF/blob/affe30ff49ceb8f09ddf13ea1513d847601ad5b0/src/host/timer.h#L43 Can the timer object be deleted while a callback...
This has been discussed a few times now (see e.g. https://github.com/microsoft/CCF/issues/1014). We should add some tests for rollback around the time a new node joins the service (for both 1tx...