Luke Frisken

Make logical sets of storage changes transactional

1

At the moment, we have logical sets of changes to the storage which can fail halfway, and leave the file system in an indeterminate/incorrect state which is difficult to recover...

Refactor round storage methods to return actions

1

The new `Round::reset()` introduced in #288 returns storage actions rather than directly mutating the storage. This easier to test and reason about. There are other methods in `Round` which could...

We already handle the situation where there are no replacement contributors, by restarting the round without the dropped contributor. But what happens when there are replacement contributors but they are...

Proposal: Refactor phase1-coordinator's coordinator.rs, coordinator_state.rs

4

Currently we are stalling on https://github.com/AleoHQ/aleo-setup/pull/256, which appears on the surface to be a bug fix, but in reality is continuing the implementation of the replacement contributors (https://github.com/AleoHQ/aleo-setup/issues/260) now that...

Better progress reporting for aggregation

3

On the universal setup the aggregation phase can take several hours, so it would be good to have some nice progress reporting

Remove pauses, perhaps by buffering network transactions

Coordinator::try_lock_chunk doesn't update the participant's locks in CoordinatorState

1

`Coordinator::try_lock_chunk()` doesn't update the participant's locks in `CoordinatorState`

Refactor Participant enum to contain structs

1

Currently there are many methods which interact with participants of only `Participant::Contributor` or `Participant::Verifier`. They are not statically type-safe (when they could be), and realistically require runtime validation of input/output....

* Daily (or more frequent) backups are important during the season. * Culling on S3 to save on storage costs means that during the off season backups are being culled...