Hi Peter,

Thanks for your comments.

>handing out xpubs makes the gap limit problem quadratic

Yes, my thinking on this is that if you're handing out xpubs you can lower the gap limit for addresses generated by those xpubs, provided you assume those addresses will be used by the same person, so there is less of a reason to expect a gap. This thread is related:
https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2022-September/020954.html

>How can we make a layer 1 address that expires

This was brought up by Sjors Provoost in relation to Silent Payments. He suggested embedding a sunset date in the address format.

>Could there be some more exotic deterministic path that doesn't split receive and change addresses

I don't follow this one. I see no reason not to split the two, and I do see potential pitfalls when you don't. Conceptually, I think receiving money twice on the same address is never good. Even if you're doing it to actively mislead people, that attempt is still leaking information that simply didn't need to be leaked.

Cheers,
Ruben

On Sat, Oct 1, 2022 at 10:57 AM Peter via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> wrote:
Hi Ruben,


I think this is an important conversation you have raised. I want to add some points for discussion.


1) handing out xpubs makes the gap limit problem quadratic.


Each customer, of a given business, on an invoice must be given a unique address or xpub but they may pay in cash or credit card or bank wire. How do we present more than 20 customers with an "invoice address" (regular address or xpub)?

(In Lightning world you give a Lightning address that uses plus addresses. Like castiron+customer1.invoice1@LSP.com


If you hand out xpubs it can be the case that you hand out a consecutive streak of 20 xpubs that are never used. Your wallet has to scan 20 xpubs and their 20 first receive addresses.



2) Whether you give the sender an address for reuse or an xpub for reuse there needs to be an expiry such that the receiver can confirm they still have the corresponding keys. How can we make a layer 1 address that expires like a PGP key where it can still be used but raises a warning to the sender?

(In Lightning we have that)


3) Could there be some more exotic deterministic path that doesn't split receive and change addresses? What is the first principle of splitting change and receive? What's wrong with an address reused exactly twice? The sender and receiver both with know what was a payment and what was change. Will it create plausible deniability about change addresses?


Satoshi original wallet concept was an ever growing key pool with a 100 address "gap". Maybe the solution to the gap limit is to add invoice functionality to wallets that manage issuing fresh addresses even without them being used and have a configurable gap limit. Is that what Btcpayserver does?


Regards

Peter Kroll

_______________________________________________
bitcoin-dev mailing list
bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev