Review FlyClient proposal for possible impact on Light client direction

[FrankSzendzielarz]

There is a "FlyClient" proposal that seems to offer an alternative approach for constrained resource clients . The document is at the this location .

A review would help us understand

• What it offers
• Should it be implemented and if so how
• What is the effect on the converged light protocols project (bringing Parity and geth in alignment for the future of PIP + LES)

[rjl493456442]

Never found it until now! Actually I read this paper several months ago.

TL;DR the main idea of this paper is: the client can verify a checkpoint of blockchain without the trust of any third party.

So that when a client joins the network, it can:

• request a checkpoint of longest blockchain;
• verify the validity of checkpoint via sample(I think it uses the similar idea of zk, based on the formula the paper constructs, the client can detect fraud with a very high chance in limited steps.);
• sync the remaining header based on the verified checkpoint;

Personally I think it's awesome. Now we use a centralized way to prove the checkpoint is valid(only foundation developers can generate checkpoints). A trustless checkpoint is always the goal we pursue