[CSL-24] Move local MPC data from Acidic storage about cardano-sl HOT 8 CLOSED

@georgeee

Local data is not to be stored in ACID

from cardano-sl.

@Gromak

Yes, we must move local data into memory state. It doesn't make much sense to do refactoring which moves local data from version as a separate task. It will be more efficient to move local data into memory state at once.

That's why I have updated this issue.

from cardano-sl.

@Gromak

And now 12h estimation is reasonable :)

from cardano-sl.

@neongreen

Looks okay to me. Where LocalData ssc (which is going to be an IORef, right?) would be stored?

from cardano-sl.

@Gromak

@neongreen good question. It should be stored on very high level, higher than Worker or Server, because it's needed for both Worker and Server.
I think we can do it in the same way as we deal with NodeState (MonadDB). I will call it MonadSscLocalData, but it's not the best name. Then sscProcessMessage and sscGetLocalPayload will require MonadSscLocalData constraint (instead of taking LocalData as argument) and we can implement MonadSscLocalData in the same place where we implement MonadDB.
Does it make sense?

from cardano-sl.

@Gromak

Btw, we also need to add something like processBlock to SscLocalDataClass, because we need to remove local data which appears in blocks.

from cardano-sl.

@Gromak

Regarding atomicity: I don't see any fatal problems. We must be careful though, because, for example, when we pass global payload to sscProcessMessage it may happen that global payload changes. But actually we just need to check local payload before adding it to block, that's all.

This part is not implemented, I have created issue about it: CSL-155.

from cardano-sl.

@Gromak

In general we have finished this issue, but there are some improvements to be done: CSL-154, CSL-155 and CSL-153.

from cardano-sl.