From rusty at rustcorp.com.au  Fri Aug 21 01:27:22 2015
From: rusty at rustcorp.com.au (Rusty Russell)
Date: Fri, 21 Aug 2015 10:57:22 +0930
Subject: [Lightning-dev] A state machine.
In-Reply-To: <CAMXsxZNRDkJVa2YFLDk_wyBigdGp-WtqwjeGD4-HDX5JkrZBMA@mail.gmail.com>
References: <87si7eiehg.fsf@rustcorp.com.au>
	<CAMXsxZNRDkJVa2YFLDk_wyBigdGp-WtqwjeGD4-HDX5JkrZBMA@mail.gmail.com>
Message-ID: <87k2spig79.fsf@rustcorp.com.au>

Pierre <pm+lists at acinq.fr> writes:
> Hello all,
>
> First, bravo for the great work on lightning ! @Rusty I'm the guy who
> recently made two dummy PR on your github project, thanks for merging them
> ;-)
>
> I've just got one remark : from OPEN_WAITING state, you seem to be assuming
> that the event BITCOIN_ANCHOR_DEPTHOK will always happen before receiving
> the other party's PKT_OPEN_COMPLETE ; but that won't necessarily be the
> case depending on each party's minDepth, right ?

It actually ignores ("defers") that packet until after the transition;
I suppressed the loops in the simplified diagram, but the code is:

	case STATE_OPEN_WAITING_OURANCHOR:
		if (input_is(input, BITCOIN_ANCHOR_DEPTHOK)) {
			set_effect(effect, send,
				   pkt_open_complete(effect, sdata));
			return STATE_OPEN_WAIT_FOR_COMPLETE_OURANCHOR;
		} else if (input_is(input, BITCOIN_ANCHOR_UNSPENT)) {
			goto anchor_unspent;
		} else if (input_is(input, PKT_OPEN_COMPLETE)) {
			/* Ignore until we've hit depth ourselves. */
			set_effect(effect, defer, input);
			return state;
                ...

> Also, can you please confirm that the following is correct in an
> Alice->Bob->Carol->Dave scenario if we look at the state of *Bob* ?

Bob has two states here; the states are per connection.

So there's Bob->Alice and Bob->Carol.

> 1) Bob is in NORMAL state
> - Bob receives update_add_htlc from Alice
> - Bob sends update_accept to Alice
>
> 2) Bob switches to WAIT_FOR_UPDATE_SIG state
> - Bob receives update_signature from Alice
>
> - (Bob sends update_add_htlc to Carol on another channel)
> - (Bob receives update_complete_htlc on another channel)
>
> 3) Bob switches to NORMAL state

So far so good.  Now the HTLC is established, so I don't understand
these transitions:

> - Bob sends update_accept to Alice
>
> 4) Bob switches to WAIT_FOR_UPDATE_SIG state
> - Bob receives update_signature from Alice
>
> 5) Bob switches to NORMAL state

Cheers,
Ruysty.