Update LayerStatus for 0.2 Consensus Changes

[avive]

Add new layer statuses to conform with consensus protocol changes. Refactor this change from api-related changes in Api 2.0 PR.

[lrettig]

I'm a bit confused about this PR. I thought we already agreed (more or less) upon an updated spec here, for API 2.0? #144 (comment)

Why not just go with this?

[avive]

this is for spacemesh 0.2 and not for 0.3 hence a separate pr and should go into 1.x and not into 2.0. It has been a while since we discussed this but as far as i remember we reviewed this together and agrees that this reflects 0.2 consensus.

[lrettig]

this reflects 0.2 consensus

Could you refresh my memory? Is there a thread I can review?

confirmed by tortoise
verified by tortoise

what's the difference between these two?

[avive]

The answer depends on possible layer states in the node because the API just need to reflect them. We need to get someone who's familiar with latest Tortoise implementation to chime in on this. It is my understanding that the node has a separate notions of verified and confirmed and also about layers it knows about which are not verifier or confirmed - pending. @noamnelke - what's the deal here?

[lrettig]

I'm quite familiar with the latest Tortoise implementation :) From the perspective of Tortoise, a layer can only have two states, unverified or verified. (You could maybe posit a third, but I don't see what we gain from it: tried to verify, and failed, and will try again later.)

You may be thinking of the case where Hare finishes for a layer which hasn't been verified by Tortoise yet, but this is already accounted for in the API.

[avive]

I'm quite familiar with the latest Tortoise implementation :) From the perspective of Tortoise, a layer can only have two states, unverified or verified. (You could maybe posit a third, but I don't see what we gain from it: tried to verify, and failed, and will try again later.)

Cool, so, are you 100% sure that there is no separate verified and confirmed tortoise layer states in the protocol 0.2? If not then we don't need this change, if yes, then we do.

You may be thinking of the case where Hare finishes for a layer which hasn't been verified by Tortoise yet, but this is already accounted for in the API.

If a layer is not yet approved by hare but known to node that it won't be reported by the 0.1 api. There can be a layer in this state known to node which will not be reported via the 0.1 api as it doesn't have a state for it.

[lrettig]

This is not necessary for SM 0.2. Going to go ahead and close this. Let me know if I've missed anything!