Quote from: sickpig on June 02, 2015, 01:25:03 PM

If memory serves 2013 hardfork was more a "dependency" bug imho. Client with ver <= 0.8.0 cant's store block bigger than 512KB due to a Berkeley db default misconfiguration. It could have happened with any other external tool used by the client.

The bug wasn't nearly so deterministic. Whether or not a particular block would trigger the bug depended on the runtime history of a particular node (when they started downloading blocks, how many orphans they had observed, etc) meaning that binary-identical versions of Bitcoin Core running on machines with identical hardware could exhibit divergent behaviour based on them connecting to different peers at different times.

Quote from: sickpig on June 02, 2015, 01:25:03 PM

In that regard what do you think about the plan to isolate the consensus code in a separete library?

Would have been a good idea to do from the beginning. I'm not sure if the existing code can be cleaned up well enough to make a meaningful difference.