pull down to refresh

AI tools were not used in this code review. With 73,057 lines of diff output, I certainly am not going to claim I did a review sufficiently detailed to catch any malicious code changes from the point of view of a Knots user

neat, neat. but AI is red-teaming everything now?!

appreciate the write-up!

reply

Ha!

When everyone else is using AI it's worth it if at least one guy does not.

reply

How do you "code" without AI nowadays?

reply

Surprisingly easily. Remove the "vibe" bit and just do the thing.

reply
172 sats \ 2 replies \ @Natalia 7 Aug

I like how calm you shared the review, without causing any heart attack:)

Imagine if the ones who actually know quietly audit the code, communicate, and fix everything in private, aren't this better for everyone?

reply

There's a huge backlog of that. The AI teams (red, black, blue and transparent ones that you can't see) are attacking this backlog.

reply
235 sats \ 1 reply \ @Kruw 7 Aug

No off by one bug in this one? #2X

reply

Ha!!!

Not that I know of. Though the potential of a new poison block attack that I mentioned would be interesting to look into further.

reply
129 sats \ 1 reply \ @optimism 7 Aug

Thanks for this, Peter!

reply

To summarize my findings, I do not believe that the BIP-110 implementation does anything particularly interesting from the point of view of someone running a non-BIP-110 node. Absent some kind of attack, given their current hash power they will most likely fork off into a separate coin with minority hash power (or even no hash power).

1 sat \ 0 replies \ @fifoofa 8 Aug -30 sats

Say what you want about the fork, a real code review is the most adult thing in this whole circus. The signaling memes got 100x the attention of the actual diff and that's exactly backwards. Spec bugs are where forks die, not Twitter arguments. Even if BIP-110 is a band-aid, at least somebody checked it's clean before slapping it on.