Juan Benet

Yep agreed. On Tue, Aug 23, 2016 at 23:19 Tangent128 [email protected] wrote: > Type (bug, feature, meta, test failure, question): feature > Area (api, commands, daemon, fuse, etc): dht >...

from https://github.com/ipfs/go-ipfs/pull/1747/files#r40535116: - [ ] consider how this plays with header from ipfs/kubo#1747 - [ ] perhaps make an option like `Gateway.Robots.AllowIndexing`, default to false. - [ ] option drives...

noting that in lists like this, reporting back tipsets at log intervals (log_2 or log_10) is a good choice. ``` ... ```

or to make it easier to calculate, can use the block epoch and zero out places, from most specific to least specific. it's a log list, but at rounded intervals....

I tend to agree. we moved this to speed things up, making the point that if we cant trust our own local repo we're in trouble already. But i think...

> bitswap saves them without any sanity checking using their hash as their name that's incorrect (or should be). all blocks **MUST** be checked before saving. > So unless the...

> I'm interested in helping with building a real adversary model because it actively overlaps with my research (We can handle 'variable' levels of required security by breaking it down...

@whyrusleeping is that captured in documentation anywhere? it should also be captured in some security notes. Poisoning or corrupting repos could be an attack vector.

- I'd also err on the side of turning it on by default and letting people turn it off for speed. there could be a list of "knobs to turn...

Well put, @BrendanBenshoof. I'm on Reasoning B all the way through. If the user wants to _decrease security to increase performance_, that's their prerogative and we can give them a...