massa icon indicating copy to clipboard operation
massa copied to clipboard

Published 20 hours ago • verified publisher icon massalabs

Change endorsements

Open qdrn opened this issue 2 years ago • 5 comments

Endorsements should satisfy the following constraints:

  • Endorsements should be created and propagated to the network by endorsers at half-slot time.
  • Endorsements should endorse the best parent in the given thread as seen by the endorser at this timestamp.
  • Endorsements created at a half-period can only be included in the block of the next period.

Side note: currently when we include the endorsements in a block we verify that they endorse the parent of the created block in this thread.

cc @sebastien-forestier @leoloco

qdrn avatar Sep 07 '22 16:09 qdrn

Current implementation

Currently, we PoS-select endorsement creators targeting the blockclique block at slot (period, thread) if any. The endorsements selected at (period, thread) are created/emitted at time genesis_timestamp + period * t0 + thread * (t0/thread_count) + t0/2.

Implementation change

We need to PoS-select endorsement creators that are allowed to be included in blocks at slot (period, thread). The endorsements selected at (period, thread) are created/emitted at time genesis_timestamp + period * t0 + thread * (t0/thread_count) - t0/2 and endorse the latest blockclique (or final) block of their thread.

damip avatar Sep 09 '22 12:09 damip

One additionnal change to be discussed: take into account the compatible endorsements in the pool to compute the cliques fitness

sebastien-forestier avatar Sep 11 '22 16:09 sebastien-forestier

One additionnal change to be discussed: take into account the compatible endorsements in the pool to compute the cliques fitness

What cases do you have in mind ? If you create a block you create the one that gives the clique with the highest fitness (I guess ?). If you create endorsements then you probably don't have the other endorsements.

qdrn avatar Sep 11 '22 20:09 qdrn

Part 1 is done here: https://github.com/massalabs/massa/pull/3004 Not merging yet (breaking)

damip avatar Sep 13 '22 13:09 damip

Waiting for feeedback from research team.

AurelienFT avatar Nov 25 '22 08:11 AurelienFT

I think we can close this one. Right ?

qdrn avatar Apr 05 '23 14:04 qdrn

take into account the compatible endorsements in the pool to compute the cliques fitness

This hasn't been discuss but could be moved to a discussion

AurelienFT avatar Apr 05 '23 14:04 AurelienFT

take into account the compatible endorsements in the pool to compute the cliques fitness

This hasn't been discuss but could be moved to a discussion

Yeah, I don't think it belongs to an issue.

qdrn avatar Apr 05 '23 15:04 qdrn