UkoeHB: The current address format doesn't have a checksum at the consensus level. A checksum is enforced by wallet software. With what you're working on, (1) is it possible to have an address checksum, (2) at what level of enforcement would it be, and (3) are there any costs to a checksum?

Consensus does not *know* the addresses, and should not.

What do you mean by checksum?

what problem are you trying to solve for?

Rucknium[m]: maybe this kind of thing is what you want? mobilecoinfoundation/mcips #4

err I guess this would be the text: github.com/mobilecoinfoundation/mci…-recoverable-transaction-history.md

<luigi1111> "what problem are you trying to..." <- Mistyping the address by one character and sending XMR to the void accidentally.

Does that happen? The address format has an internal checksum

UkoeHB: Does an internal checksum mean that a tx with a mistyped address won't be mined? Or that it would be impossible to construct, anyway, even with bad wallet software?

It a tx is created, it will be mined (as long as valid, miners aren't jerks, etc).

The checksum's at least 32 bits though. Very unlikely to match for a random change.

What I'm trying to understand is if the employed cryptography prevents a mistyped address through a checksum or if what is being relied upon is that the wallet software has been written well.

The latter.

For example, the GUI wallet is fine, but would an alternative wallet implementation (there are many, of course), skip the checksum.

I see. Thanks

Your software would have to be pretty shit if it's ignoring the checksum.

Have we checked if the major alternative wallets implement the checksum?

Define "we".

I think jberman uncovered evidence that some wallets are using really old mixin selection algorithms, which is why I think it's plausible that a wallet could skip the checksum.

s/mixin/decoy/g ...

<hyc> "s/mixin/decoy/g ..." <- Oops. Right. Old habits die hard. Should have said "decoy selection algorithm".

Thanks Rucknium and isthmus @isthmus_mpkt:matrix.org for the detailed updates on your respective work. I'm looking forward to reading more, and the mitigation discussions. If you don't mind, I will post your updates as an addendum to the meeting logs because they happened right after and relate to agenda items

Thanks @carrington[m] 👍

carrington: Sounds great. Thank you for doing the legwork.

the checksum has been the same since the beginning

so old doesn't really matter (they might not recognize new address types, but should error in that case)

if (!valid(address)) address = MYADDRESS;

/* fallback to avoid user error */

very good