Sadly, I am undecided there are nice solutions to those questions proper now, however I am going to attempt to clarify how issues work at this time.
If doing a 2nd CPFP do you have to enhance the unique tx, or enhance the final enhance?
I believe it is dependent upon what you are making an attempt to perform. In the event you’re comfortable with the primary CPFP try taking some time to verify and also you simply need the very first transaction to verify quickly, then you definitely may need to simply CPFP the unique transaction once more (assuming you’ve gotten one other output to make use of?), so that you simply solely are paying for 2 transactions to rapidly verify, fairly than 3.
If as an alternative you want the primary CPFP try to verify rapidly as properly, then you could possibly spend certainly one of its outputs and attempt to price bump each the primary and second transactions whenever you create this third one. Transactions are chosen for mining based mostly on their “ancestor feerate” rating, which is the entire feerate of a transaction together with all of the unconfirmed mother and father that have to be confirmed as properly to ensure that the transaction to be legitimate to be mined. So this is able to do precisely what you need, I believe.
Alternatively, you could possibly additionally attempt to RBF the primary two transactions on the similar time, by issuing a transaction that conflicts with the primary one and contains the outputs you need to embody from each these transactions. Whether or not that is extra economical will rely on what different transactions are within the mempool that spend outputs of those two transactions (see beneath).
What occurs to CPFP txs which might be already within the mempool and the unique tx will get RBF’d?
Any transactions that rely on the unique can be faraway from the mempool when the unique transaction is RBF’d. Usually, RBF is extra environment friendly (decrease required charges) than CPFP if the objective is to fee-bump. Nevertheless, for anti-DoS causes, if there are baby transactions within the mempool, then RBF’s of the guardian are required to not solely pay for the guardian to be evicted (the next price and better feerate than the guardian), however they’re additionally required to pay for all the youngsters being evicted, too (so the brand new transaction will need to have the next price than that paid by all transactions which might be faraway from the mempool). That is described in BIP 125.
So, in case you have a transaction that another person has already chained a CPFP onto, then it is perhaps dearer so that you can RBF your authentic transaction than it’s so that you can chain your individual CPFP transaction, as a result of the RBF wouldn’t simply pay for changing the guardian but additionally the kid that another person created.
Additionally, a limitation of the present mining algorithm is that in a state of affairs the place a guardian transaction has two completely different kids making an attempt to CPFP it (ie “sibling transactions” that every solely rely on the guardian, and never one another), the precedence with which the guardian can be mined can be on the larger of the 2 ancestor feerates of these kids — regardless that it is potential that one of the best feerate could be the one that features all kids taken collectively.
Would it not be more cost effective to carry out a number of CPFP txs, or do you have to contemplate changing your authentic CPFP tx with RBF?
Within the absence of different transactions that may spend outputs of your authentic transaction — which may result in RBF “pinning”, the place a transaction is tough to interchange as a result of guidelines requiring replacements to pay extra complete price than all transactions that must be eliminated — it ought to all the time be cheaper to RBF than CPFP, as a result of it’s a must to pay for extra complete bytes of transaction knowledge to be mined whenever you generate new transactions than whenever you substitute a transaction.
Nevertheless, RBF could be tough as properly, as a result of if you happen to RBF a transaction repeatedly it’s a must to watch out about the way you handle your inputs,
in order that if you’re producing a number of replacements for a transaction, then each ought to battle with everybody that you simply signed earlier than, or else you may find yourself issuing a cost twice. So some thought is required in designing how a given utility or pockets is designed to guarantee that that is used correctly.
Sooner or later, the pinning points associated to RBF could also be mitigated by new coverage guidelines, such because the “v3” transaction relay coverage that has been proposed, which might restrict RBF pinning by limiting the dimensions and variety of baby transactions that transactions opting-in to that coverage are permitted. Nevertheless, that is not one thing that has been deployed but.