Post

eg, everything that does handle resolution, or OAuth client flows, or fetches identity info or data w/o validation, or service proxying, will need some form of operational update or mitigation.

1 repost38 likes

I'm sort of wishing more of the atproto identity code I have written was doing real PLC validation, instead of just fetching the DID doc...

i thought about that too but that requires pulling all N DID docs that have ever existed for that user, yeah?

I supported it in atproto-crates for a hot minute, and removed it recently for the current release candidate. I think it is a sound and reasonable DID method. that we should be more widely supporting and using.

pulling the full PLC audit log. which usually isn't that big, and probably compresses well, but yeah there are tail cases. replicas help with load on the directory. a lot of services (clients) don't do many resolutions. ones that do will probably have good caching, maybe local validating replica

reference read-replica validating ops is the very best thing about it imo. only helps here if you access it not-over-public-tls but that's very possible if you self-host it. (but also, read-replica perf when fetching full identity audit logs is also very good)