Transcript for #bitcoin-dev 2017/02/25

17:24 tommygunner evening
17:24 tommygunner what happens if i use replace-by-fee but a transaction with the lower fee is confirmed first
17:25 tommygunner does the one/the ones with higher fee clear from the mempool automagically or do they hang around
17:33 arubi tommygunner, replace by fee also means that you're double spending the low fee transaction, so if it does get confirmed eventually, the higher fee one is no longer valid
17:34 arubi they're both valid before one is in a block, but will not appear together in someone's mempool. only one of them will
17:34 tommygunner i think the problem is with the receiving end's implementation
17:35 arubi shouldn't be? they can't live together on the same mempool, and if the "other one" gets in a block, then the one you have will be dropped from your mempool
17:36 tommygunner the old ones are still listed on their website as unconfirmed transactions
17:36 tommygunner i think i will mail them and ask about replace-by-fee
17:37 arubi if they are an actual double spend of the one that did get confirmed (one of the inputs is double spent), then yea, not so useful to have the unconfirmed one listed (except for historical use I guess)
17:37 arubi s/they are/it is
17:38 tommygunner that would mean the only way to resolve it is wait for a long time until they drop it
17:39 arubi I don't understand, did the original transaction also pay to them?
17:39 arubi I guess it did if you used RBF
17:40 arubi anyway, yea, weird. if the original one is in a block, then why track this one? that's what happens when services track txid's instead of supplying the customer with a unique address :)
17:40 tommygunner its a little bit complicated because it was someone else that did it
17:40 arubi it doesn't really matter who pays the address
17:41 tommygunner and the first time it wasnt really a RBF but manually double spending
17:41 tommygunner with a higher fee
17:41 arubi oh, so are you sure you paid to their address in both transactions?
17:42 tommygunner yeah
17:42 arubi wonder if they used the unconfirmed funds in their own transactions and now have some chain that will never confirm
17:42 tommygunner that should not have happened, the fee increase should have been done by sending an output to your own address and not to them again
17:43 arubi the fee increase can be done any way you want
17:43 arubi the important thing is to actually double spend one of the inputs so that you don't have two independent valid transactions
17:44 tommygunner the way i did it before RBF was if there was an output in the unconfirmed transaction that i have control over, just send that to myself again with a high enough fee
17:45 arubi that's CPFP
17:45 tommygunner yeah the guy i told to do that accidentally sent it to them again
17:46 tommygunner so its a little bit of a fucked up situation
17:46 arubi then there are two independent valid transactions, it's neither RBF or CPFP :)
17:46 arubi unless you mean he used the unconfirmed output to send it to them again?
17:47 arubi (and not just "send again" which won't use unconfirmed outputs by default)
17:47 tommygunner yeah the unconfirmed output to them again
17:47 tommygunner a nice figure 8
17:48 arubi and you say that one got confirmed?
17:48 arubi the child has to confirm before the parent (can be on the same block, but physically before it)
17:48 arubi er, after the parent
17:48 arubi so, the low fee one has to be confirmed as well
17:48 tommygunner the CPFP was a RBF
17:49 tommygunner or something weird
17:49 arubi makes little sense :)
17:49 tommygunner obviously the first transaction is confirmed because without that none of the RBF could be
17:50 arubi nope, in RBF only one transaction confirms
17:50 arubi with CPFP, both confirm
17:50 arubi both, or whatever number in the chain. could also be that the parent is confirmed without the child
17:50 arubi really I think you just paid twice :)
18:06 tommygunner yes but i can withdraw again, or rather he
18:06 tommygunner but only if the transactions are all confirmed
18:10 arubi sounds like your problem isn't very related to bitcoin development, try emailing that service
18:10 tommygunner yeah i know, i just wanted some general info about rbf
18:11 tommygunner thanks for chatting, heh
19:05 arubi oh np