Daniel Knopik
Daniel Knopik
I spent some time thinking about this. Pulling the parallel state cache down into the `store` does not feel quite right to me, even if it only stores slots or...
Awesome, I'll check it out and experiment a bit and get back to you with any findings! :)
Ready for review - Happy to receive feedback, especially regarding the points listed in "Additional Info".
> So the error happens only with #4575 or is fixed with #4575? It happens both with and without #4575. I mentioned #4575 because @michaelsproul suggested that the error might...
I'll take a look at this
I investigated this a bit. 1. I believe the optimisation to load the state at the beginning of the previous sync committee period is not really the problem, as the...
>Something else that could be causing the discrepancy is this: https://github.com/sigp/lighthouse/issues/4929 Unfortunately, my fixes at https://github.com/sigp/lighthouse/pull/5101 did not make the adjusted tests in this PR pass. However, as @zack-scott has...
Regarding the expected behaviour: In the phase0 spec, the proposer is rewarded at epoch boundaries in `get_inclusion_delay_deltas`. There, slashed attesters are skipped as we iterate over the return values of...
>I guess a hybrid approach could be to retain the `shuffling_cache` as a simpler LRU, which relies on the de-dupe at the store level for parallel requests? Hmm. I think...
This is tagged as "waiting-on-author". How do we want to proceed with it?