Fork choice update for 0.9.1 - Spec PR 1465 #573
Conversation
Codecov Report
@@ Coverage Diff @@
## 0.9.x #573 +/- ##
==========================================
- Coverage 65.56% 65.49% -0.08%
==========================================
Files 198 198
Lines 3560 3599 +39
Branches 325 338 +13
==========================================
+ Hits 2334 2357 +23
- Misses 1129 1138 +9
- Partials 97 104 +7 |
There was a problem hiding this comment.
It looks good. The only difference I see with the spec is that they save the best justified root even if shouldUpdateJustifiedCheckpoint returns false. They then can set the justified root based on this, when a new epoch rolls around (based on clock time).
Imagine a scenario where the fork choice calculates that there is a better justified root, but it's already 3/4 thru the epoch, so it isn't immediately set. Then the next many blocks are left empty by validators that are offline/etc. The spec implementation will set the better justified root when the next epoch comes around, even if no blocks are around to trigger the setting of the justified root.
All in all, we should probably start a setInterval in the start method with the interval being triggered at the start of each new epoch, fulfilling the functionality of on_tick of the spec.
Goal