[consensus] Inferred Certification#3647
Conversation
Deploying with
|
| Status | Name | Latest Commit | Updated (UTC) |
|---|---|---|---|
| ✅ Deployment successful! View logs |
commonware-mcp | 63753df | Apr 21 2026, 10:02 PM |
|
I think yes we should emit finalization vote. With current code finalization path does not require a view to be explicitly verified (it just reqiures the view to not be already finalized/nullified locally, and to have natorization+certificateation for the view). So if inference marks a view as certified, emitting finalization certification seems consistent and helps liveness. I also just thought to call verify fn on inferred views but try_verify() is scoped only for current views not for ancestors and we deviate from the core problem if we try to adjust the scope of it. Also I skimmed your implementation and would love to understand your approach and thought process. |
Codecov Report❌ Patch coverage is @@ Coverage Diff @@
## main #3647 +/- ##
==========================================
- Coverage 95.88% 95.86% -0.03%
==========================================
Files 441 442 +1
Lines 172366 171965 -401
Branches 4006 4016 +10
==========================================
- Hits 165276 164853 -423
- Misses 5824 5845 +21
- Partials 1266 1267 +1
... and 39 files with indirect coverage changes Continue to review full report in Codecov by Sentry.
🚀 New features to boost your workflow:
|
2 participants
Related: #3433
Open question: should we emit a finalize vote if we infer certification (we only emit a notarize vote if we explicitly verify but the only way that is cancelled is with a notarization...with this, a future notarization can cause us to infer so there is a chance we don't form finalizations)?