Good morning Karl-Johan Alm,

To clarify:

Nothing prevents a miner from completely ignoring nSequence when putting 
transactions in blocks.

Unconfirmed transactions are, by definition, not recorded in blocks.  So if 
there is a transaction 0xFFFFFFF nSequence and fee 1000 satoshi, and another 
conflicting transaction 0xFFFFFFF nSequence and fee 100000000 satoshi, miners 
can include the latter one even if the first one came to their knowledge first, 
regardless nSequence.

Thus, in the end "full replace-by-fee", where nSequence is IGNORED for purposes 
of replace-by-fee, is expected to become the norm, and we should really be 
designing our wallets and so on so that we only trust transactions that have 

The "nSequence=0xFFFFFFFF means opt-OUT of RBF" convention is only followed by 
fullnodes running standard bitcoind.  Nothing prevents miners from running 
patched bitcoind that ignores this rule, and connecting with similar peers who 
also ignore this rule.


​Sent with ProtonMail Secure Email.​

‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐

On April 11, 2018 5:37 PM, Peter Todd via bitcoin-dev 
<> wrote:

> On Wed, Apr 11, 2018 at 05:10:43PM +0900, Karl-Johan Alm wrote:
> > On Wed, Apr 11, 2018 at 4:52 PM, Peter Todd wrote:
> > 
> > > Or via full replace-by-fee, which appears to be used by a significant 
> > > minority
> > > 
> > > of miners:
> > 
> > I was of the impression that final transactions (sequence=0xffffffff)
> > 
> > cannot be RBF'd.
> My full-replace-by-fee tree ignores that. It also does preferential peering to
> ensure it's well connected with likewise peers, and thus the whole network.
> ---------------------------------------------------------------------------------------------------------------------------------------------------------------
> 'peter'[:-1]
> bitcoin-dev mailing list

bitcoin-dev mailing list

Reply via email to