Skip to content

Comment on Why SpiderOak doesn't de-duplicate data across usersparent

Comments

... and yet SHA1 collisions have happened. According to NIST, "Federal agencies should stop using SHA-1 for...applications that require collision resistance as soon as practical"

Regrettably, we don't actually have any truly ideal hash functions yet.

Not that any of the rest of this argument makes much sense, but SpiderOak (says it) uses SHA256. There are no induced collisions on the horizon for SHA256, and there won't be accidental ones in this data set.

I did a Google search for a SHA-1 collision and I thought nobody had found a collision yet, they had just found a collision attack in the 2^50-2^60 range.

Of course, turning a collision into an exploit in this case would be challenging. (Preimage attacks ⊂ collision attacks.)

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.