From rusty at rustcorp.com.au  Thu Mar 17 04:51:08 2016
From: rusty at rustcorp.com.au (Rusty Russell)
Date: Thu, 17 Mar 2016 15:21:08 +1030
Subject: [Lightning-dev] Payment and Refund Stuck
In-Reply-To: <20160316092744.GA20196@sapphire.erisian.com.au>
References: <20160316092744.GA20196@sapphire.erisian.com.au>
Message-ID: <87fuvpn1vn.fsf@rustcorp.com.au>

Anthony Towns <aj at erisian.com.au> writes:
> Hello lightning implementors!
>
> (Posted to lightning-dev as well, hopefully that's fine with everyone)
>
> Matsjj set up a repo a while ago for collaborating on documenting the
> lightning protocol and gave a bunch of us admin access:
>
>  - https://github.com/lightning-core/lightning
>
>  """I think it is a very good idea to carry all the different
>     implementations to one spot. I think such a repository is best suited
>     for it."""
>
>  - https://github.com/lightning-core/lightning/pull/4#issuecomment-151778273
>
> I think the layout is a bit too nested (which makes sense coming from
> the Java implementor, I guess? :) and I think it'd be good to have a
> simple way to decide how to move ideas into/through the repo, without
> making it a place where there's any point trying to politicise proposals.
> (It's kind of telling that something's wrong when even matsjj hasn't
> been able to get his pull requests accepted :)
>
> Anyway I've taken Rusty's couple of BOLT proposals as well as the shachain
> design txt and rearranged them in a way I think makes sense:
>
>  - https://github.com/ajtowns/lightning-core/tree/rusty
>
> What do you guys think?
>
> I hear there's totally going to be released code out by
> northern-hemisphere summer, so documenting the protocols/standards is
> only going to get more pressing...
>
> My inclination is to add:
>
>  - matsjj's pull requests [0]
>  - the anonymous "R", via private key reveal stuff [1]
>  - Joseph's 2-of-3 Instant Escrow [2]
>
> as additional "early drafts". Probably things from Rusty's "Deployable
> Lightning" paper (like the HTLC scripts), or the "elkrem" scheme, or
> Nicholas Dorier's "backward deterministic [revocation] Value" stuff
> would be good too...

FWIW, my rough plan was BOLT #3 was going to be the transaction formats,
and BOLT #4 the onion and failure message formats (as portended in BOLT
#2's references).

Cheers,
Rusty.