Monotonicity of EB inclusion in RBs?

[bwbush]

I'm running the Haskell simulation with the default configuration and see that an EB generated in slot 80 is placed in an RB before the EB generated at slot 60 is. Is this a consequence of "freshest first"?

[Saizan]

It's not because of the freshest-first diffusion config per se. I believe at some point the Short-Leios google doc specified that nodes should include the "freshest certified EB" when generating an RB, however I can't find this in the document anymore, nor see the version history.

@pagio in Short-Leios, how should an honest node choose the EB to include in an RB among the ones still eligible?

[bwbush]

Here is the plot showing when the RBs were generated. One can see that two EBs were ready to be included in RBs after RB production had paused.

[pagio]

@Saizan The obvious way given the current formulation of things is to put in the RB the oldest EB that is still valid according to the EB freshness rule (EBs cannot be older than r_EB slots from the RB generation slot). Sorry for the ambiguity here, in Full Leios this is not really an issue.

[SupernaviX]

The rust chooses the EB with the most TXs, to maximize throughput. I don't think haskell has equivalent logic, since IBs are all identical, but maybe choosing the EB with the most IBs would be similar?