Jeffro

Yes, there would be 0 CPU performance improvement, but if we only save enough bytes from the intermediate hash to bind the contents of the block stronger than its difficulty...

Sorry, I guess I jumped ahead a bit since it wouldn't create the same hashes as the current RandomX proposal, it would require a Blake2B hash of this reduced hash,...

> The history of secure hash algorithms demonstrates fairly conclusively that you want a wide safety margin because things can and do often go wrong I would tend to agree,...

Yes everything you said @tevador is correct; we would have to double the size of the buffer to prevent collision attacks. Along with @SChernykh's point that since the hashing blobs...

Is there any way to test this against a Loongarch system?

@iamamyth AFAIK the "random" module doesn't have a way to randomly sample from a gamma distribution  Dec 14, 2023, 13:23 by ***@***.***: > > > > > @iamamyth> commented on...

Okay the reference code now doesn't select locked outputs and also doesn't use `numpy`. I'll do some KS tests soon to test if it matches the behavior of `wallet2`. The...

FWIW, I wrote a similar change in PR #9135: the caller can provide a hard fork version for which it is known that the transaction passes non-input consensus checks. If...

> Have you run this patch and validated the entire blockchain? No, not yet, but that's a good idea. > Were these types allowed at any point? If you look...

You can run this python script on a machine running `monerod` locally to convince yourself that the script/scripthash vin/vout types never appear on-chain: ```Python import bisect import json import random...