Stats on compact block reconstructions

[Crypt-iQ]

@0xB10C

limit the prefill amount to something like 10kB worth of transactions as per BIP152 implementation note #5. I think this is useful to avoid wasting too much bandwidth if a node does a high-bandwidth announcement but, for some reason, prefills a lot of well-known transactions in the announcement

One point brought up by sipa here in a semi-related thread ([WIP] p2p: Add random txn's from mempool to GETBLOCKTXN by davidgumberg · Pull Request #27086 · bitcoin/bitcoin · GitHub) is that the number of TCP packets sent over could increase if we’re making the CMPCTBLOCK message larger with prefilledtxns. I think that is maybe one downside to prefilling transactions. Perhaps it’s possible to prefill transactions up to a certain total message size limit specifically for compact blocks?

EDIT: His point was actually about the GETBLOCKTXN causing more round trips, but the same thing applies.