It was about 18 months back, with gitea, I was starting some new projects and went sha256 because I’m a nerd that adopts things early. I could do basic git stuff, but I was stunned by how much didn’t work. A lot of actions, maybe even the entire action runner itself wouldn’t/couldn’t/didn’t work. It’s a deep cutting change and an expensive one that doesn’t have a new feature value. I abandoned that effort and rebuilt my repos.
Intentional attacks and collisions aside, what do you say to the fact that Sha1 had a lifespan at design time? NIST has issued retreating usage guidelines for the last 15 years and they themselves say it should be completely phased out by 2030. It just seems like good hygiene to switch it out. I agree, it will have a long tail and be a big bowl of suck, especially for the not dead but rarely touched code.
I appreciate the intent of “modern” tools with no switches, no ways for devs to make bad security choices because the hash, cipher and encryption mode are fixed vs all the crazy looking asn.1 stuff in tls and openpgp to enable n-degrees of configuration but this very issue is the counter example.
Intentional attacks and collisions aside, what do you say to the fact that Sha1 had a lifespan at design time? NIST has issued retreating usage guidelines for the last 15 years and they themselves say it should be completely phased out by 2030. It just seems like good hygiene to switch it out. I agree, it will have a long tail and be a big bowl of suck, especially for the not dead but rarely touched code.
I appreciate the intent of “modern” tools with no switches, no ways for devs to make bad security choices because the hash, cipher and encryption mode are fixed vs all the crazy looking asn.1 stuff in tls and openpgp to enable n-degrees of configuration but this very issue is the counter example.