search

coniks-sys / coniks-go

A CONIKS implementation in Golang
http://coniks.org
Other
116 stars 30 forks source link

Move policies to protocol #121

Closed arlolra closed 7 years ago

arlolra commented 7 years ago

See #120

This needs a sanity check.

vqhuy commented 7 years ago

I'm not sure how do you think about this, but according to #50, and this statement

it makes the PAD more general, perhaps to be reused in other contexts

I imagine that merkletree would have a Hasher interface that exposes a HashID. Does this mean merkletree should control what is the value of HashID being included in AssociatedData (even without #50, I guess)?

In the other hand, if we go this route, Hasher and HashID are probably parts of protocol. What I mean is HashID is definitely a part and being used in PAD but now it's still being used in PAD but a part of protocol. (Sorry for my bad expression). It feels like to have a conflict here.

Back to #120, I think your main concern is storing many copies of vrfPrivateKey in memory / on disk, so should we just remove vrfPrivateKey out of Policies and keep Policies in merkletree (We could reuse a part of your code here). Does this make sense?

arlolra commented 7 years ago

I started working on this a bit in c355ab1b.

It feels like to have a conflict here.

I see what you're saying, but it's not clear that everyone who uses the PAD would necessarily want to include the hash id as part of the chain.

We can definitely put this on the agenda for tomorrow :)

arlolra commented 7 years ago

Ok, implemented AssociatedData as an interface. Let me know what you think. We can discuss tomorrow as well, if preferred.

arlolra commented 7 years ago

Merged in 29020d03.