Transaction Replacement by Fee

Hello,

Where can I find the rules on transaction replacement?

For instance what are the valid ranges for the sequence values in transaction inputs? Normally I set this value to MAX_UINT32, and the locktime to 0.

Can I change the outputs? Can I add more inputs to pay a higher fee?

Just looking for clarity on this aspect of Bitcoin. Any resources would be much appreciated.

Thanks.

BIP125 is the standard way to signal: https://github.com/bitcoin/bips/blob/master/bip-0125.mediawiki

Should explain everything you need.

On Thu, Jan 12, 2017 at 9:02 AM, Police Terror via bitcoin-dev < bitcoin-dev@lists•linuxfoundation•org> wrote:

> Hello,
>
> Where can I find the rules on transaction replacement?
>
> For instance what are the valid ranges for the sequence values in
> transaction inputs? Normally I set this value to MAX_UINT32, and the
> locktime to 0.
>
> Can I change the outputs? Can I add more inputs to pay a higher fee?
>
> Just looking for clarity on this aspect of Bitcoin. Any resources would
> be much appreciated.
>
> Thanks.
> _______________________________________________
> bitcoin-dev mailing list
> bitcoin-dev@lists•linuxfoundation•org
> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>

On Thu, Jan 12, 2017 at 09:13:23AM -0500, Greg Sanders via bitcoin-dev wrote:

> BIP125 is the standard way to signal:
> https://github.com/bitcoin/bips/blob/master/bip-0125.mediawiki
>
> Should explain everything you need.

Additionally some miners mine full replace-by-fee, which has no limitations on nSequence. My implementation (for v0.13.2) is here:

    https://github.com/petertodd/bitcoin/tree/replace-by-fee-v0.13.2

and is identical to Bitcoin Core modulo the nSequence stuff being removed, and a special service bit added to allow full-rbf nodes to preferentially peer with each other to make sure replacement transactions get propagated.

In practice full-RBF works fairly well, so while it's even faster to use the nSequence signalling specified in BIP-125, doing so is not mandatory so long as you can et your replacement transaction to a full-RBF node.

Hi, recently been trying to get RBF working on a multisig input. I set the nSequence to 0, but script didn't verify (used python-bitcoinlib). Should it work for this type of transaction? I am using the SignatureHash(...) method of signing, not rpc.signrawtransaction(...).

Thanks

On Thu, Jan 12, 2017 at 7:58 PM, Peter Todd via bitcoin-dev <bitcoin-dev@lists•linuxfoundation•org> wrote:

> On Thu, Jan 12, 2017 at 09:13:23AM -0500, Greg Sanders via bitcoin-dev wrote:
>> BIP125 is the standard way to signal:
>> https://github.com/bitcoin/bips/blob/master/bip-0125.mediawiki
>>
>> Should explain everything you need.
>
> Additionally some miners mine full replace-by-fee, which has no limitations on
> nSequence. My implementation (for v0.13.2) is here:
>
>     https://github.com/petertodd/bitcoin/tree/replace-by-fee-v0.13.2
>
> and is identical to Bitcoin Core modulo the nSequence stuff being removed, and
> a special service bit added to allow full-rbf nodes to preferentially peer with
> each other to make sure replacement transactions get propagated.
>
> In practice full-RBF works fairly well, so while it's even faster to use the
> nSequence signalling specified in BIP-125, doing so is not mandatory so long as
> you can et your replacement transaction to a full-RBF node.
>
> --
> https://petertodd.org 'peter'[:-1]@petertodd.org
>
> _______________________________________________
> bitcoin-dev mailing list
> bitcoin-dev@lists•linuxfoundation•org
> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>