[Lightning-dev] DRAFT: interactive tx construction protocol

lisa neigut niftynei at gmail.com
Tue Jan 28 01:51:28 UTC 2020


Some of the feedback I received from the check-in for the dual-funding
proposal this past Monday was along the lines that we look at simplifying
for breaking it into smaller, more manageable chunks.

The biggest piece of the dual-funding protocol update is definitely the
move from a single peer constructing a transaction to two participants.
We're also going to likely want to reuse this portion of the protocol for
batched closings and splicing. To that extent, it seemed useful to
highlight it in a separate email.

This is a change from the existing proposal in the dual-funding PR #524
<https://github.com/lightningnetwork/lightning-rfc/pull/524> -- it allows
for the removal of inputs and outputs.

The set of messages are as follows.


Note that the 'initiation' of this protocol will be different depending on
the case of the transaction (open, close or splice):

1. type:   440 `tx_add_input`

2. data:

    * [`32*byte`:`channel_identifier`]

    * [`u64`:`sats`]

    * [`sha256`:`prevtx_txid`]

    * [`u32`:`prevtx_vout`]

    * [`u16`:`prevtx_scriptpubkey_len`]

    * [`prevtx_scriptpubkey_len*byte`:`prevtx_scriptpubkey`]

    * [`u16`:`max_witness_len`]

    * [`u16`:`scriptlen`]

    * [`scriptlen*byte`:`script`]

    * [`byte`:`signal_rbf`]

1. type: 442 `tx_add_output`

2. data:

    * [`32*byte`:`channel_identifier`]

    * [`u64`:`sats`]

    * [`u16`:`scriptlen`]

    * [`scriptlen*byte`:`script`]

1. type: 444 `tx_remove_input`

2. data:

    * [`32*byte`:`channel_identifier`]

    * [`sha256`:`prevtx_txid`]

    * [`u32`:`prevtx_vout`]

1. type: 446 `tx_remove_output`

2. data:

    * [`32*byte`:`channel_identifier`]

    * [`u64`:`sats`]

    * [`u16`:`scriptlen`]

    * [`scriptlen*byte`:`script`]

1. type: 448 `tx_complete`

2. data:

    * [`32*byte`:`channel_identifier`]

    * [`u16`:`num_inputs`]

    * [`u16`:`num_outputs`]

1. type:  448 `tx_sigs`

2. data:

    * [`channel_id`:`channel_identifier`]

    * [`u16`:`num_witnesses`]

    * [`num_witnesses*witness_stack`:`witness_stack`]

1. subtype: `witness_stack`

2. data:

    * [`sha256`:`prevtx_txid`]

    * [`u32`:`prevtx_vout`]

    * [`u16`:`num_input_witness`]

    * [`num_input_witness*witness_element`:`witness_element`]

1. subtype: `witness_element`

2. data:

    * [`u16`:`len`]

    * [`len*byte`:`witness`]



## General Notes

- Validity of inputs/outputs is not checked until both peers have sent
consecutive `tx_complete`  messages.

- Duplicate inputs or outputs is a protocol error.

- Feerate is set by the initiator, or in the case of a closing transaction,
negotiated before the transaction construction is initiated.

- Every peer pays fees for the inputs + outputs they contribute, plus
enough to cover the maximum estimate of their witnesses. Overpayment of
fees is permissible.

- Initiator is responsible for contributing the output/input in question,
i.e. the

  funding output in the case of an opening, or the funding input in the
case of a close.

  (This means that the opener will pay for the opening output). In the case
of a splice,

  the initiator of the splice pays for the funding tx's inclusion as an
input and the

  new 'funding tx' output.

- Any contributor may signal that their input is RBF'able. The nSequence
for this input should be set to 0xFEFF FFFF, 0xFFFFFFFF otherwise.

- The initiating peer is understood to be paying the fee for the shared
transaction fields (nVersion [4], segwit marker + flag [2], input + output
counts [2-18], witness count [1-9], nLocktime [4]; total [13-40bytes])

- Inputs MUST be segwit compatible (PW* or P2SH-PW*)

- All output scripts must be standard

- nLocktime is always set to 0x00000000.

- The `num_inputs` and `num_outputs` in `tx_complete` is a count of that
peer’s final input and output contributions, net any removals.

- Either peer may add or remove inputs and outputs until both peers have
successfully

  exchanged a `tx_complete` message in succession.

- Either peer may only add or remove their own input or output.

- In the case that a `tx_complete` agreement cannot be reached, either peer
may

  fail the channel or open protocol (whatever is reasonable for the
particular case)

  - In the case of a splice, this would be a soft error (channel returns to
normal operation until

    otherwise failed or closed.)

  - In the case of an open, this would be a failure to open the channel.

  - In the case of a close, a failed collaborative close would result in an
error and a unilateral close.

### Considering the Simple Open case (2 parties)

- Both peers signal `opt_dual_fund`

- Opener initiates a channel open with `open_channel2` message, indicating
the feerate for the opening transaction

- Accepter signals acceptance of channel open as proposed, including
proposed feerate, via `accept_channel2`

- Opener sends `tx_add_output`, with the funding output for the sum of both
peer’s funding_amount

- Opener sends `tx_add_input` for each input the wish to add to the funding
transaction

- Opener sends `tx_add_output` for their change

- Opener sends `tx_complete`

- Accepter sends `tx_add_input` for each input they wish to add to the
funding transaction

- Accepter sends `tx_add_output` for their change.

- Accepter sends `tx_complete`

- Opener sends `tx_complete`

- Opener and accepter exchange commitment signatures; etc.

### Considering the Splice case:

- Both peers signal `opt_splice_ok`

- One peer initiates a splice, also signaling the feerate for the
transaction. Exact protocol unspecified herein.

- Initiator sends `tx_add_input` with the original funding output

- Initiator sends `tx_add_output` with the new, post-splice funding output

- Initiator sends `tx_add_input/output` as needed to add all desired inputs
+ outputs

- Initiator sends `tx_complete`

- Peer sends `tx_add_input/output` as needed to add all desired inputs +
outputs

- Initiator sends `tx_complete`

- Peer sends `tx_complete`

- Initiator + peer exchange commitment signatures, etc.

### Considering the Close case:

- Both peers signal `opt_collaborative_close` in their `node_announcement`.

- A peer initiates a close sending a `shutdown`, as per usual.

- A feerate is negotiated. Out of band for this particular portion of the
protocol.

-The closing initiator (peer which first sent `shutdown`), sends
`tx_add_input` to spend the funding output and `tx_add_output` to add their
output for the channel closure.

- The peer responds with `tx_add_output`, adding their output to the close
transaction.

- If `option_upfront_shutdown_script` is flagged but no such output with a
value at or within a reasonable feerate gap of the peer's funding output is
present, then the peer must fail the channel.


## Updating a collaborative transaction with RBF:

- If any input is flagged as RBF’able, then the transaction is considered
eligible for RBF

- RBF can be initiated by either party, and serves as an initiation for
another round of transaction composition, as outlined above.

- Note that this section has been cribbed and re-purposed from the original
RBF proposal for splicing, see
https://lists.linuxfoundation.org/pipermail/lightning-dev/2018-November/001621.html

1. type: 45 (`init_rbf`) (`option_collaborative_rbf`)

2. data:

   * [`32`:`channel_id`]

   * [`4`:`fee_step`]

Each `fee_step` adds 1/4 (rounded down) to the initial

transaction feerate. eg. if the initial feerate was 512 satoshis per
kiloweight, `fee_step` 1

is  512 + 512 / 4 = 640, `fee_step` 2 is 640 + 640 / 4 = 800.

The sender:

  - MUST set `fee_step` greater than zero and greater than any prior
`fee_step`.

The recipient:

  - if the new fee exceeds the sender's current balance minus reserve

    after it is applied to the splice transaction:

    - MUST error.



NOTES:

1. 1/4 is a reasonable minimal RBF, but as each one requires more

   tracking by the recipient, serves to limit the number you can create.

2. Rule 4 of BIP125 requires a feerate increase to at least surpass the
minimum transaction relay setting. Ratcheting by 25% should satisfy this
requirement

3. An additional rule will be added to the checks of an RBF transaction
that it must include at least one identical, replaceable input as the
original transaction.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxfoundation.org/pipermail/lightning-dev/attachments/20200127/ca2e504d/attachment-0001.html>


More information about the Lightning-dev mailing list