commit to make a 2x RBFr on bitcoin core
Heh, there's actually an important oversight in that specific commit that I fixed in a follow up.
Homework problem: try to spot the bug!
Extra credit: what additional problem does the fix introduce? (This one, IIUC, is not very important)
reply
peter you mean to tell me you didn't go back and fix up the commit where you introduced the error?? 😱
reply
Well, these commits come from my libre-relay-v25.0 branch that people are actually running on real nodes via git pull && make.
If I make a "proper" release I'll probably squash these commits so they make more sense in isolation.
Notice how I'm also missing the test cases that I added later.
reply
Would CFeeRate original_feerate(mi->GetModifiedFee(), mi->GetTxSize()); not want to use ModifiedFee? since it's fetching the original fee rate?
reply
Yes, I originally used GetModifiedFee, which wasn't correct as the fee was modified with, IIRC, ancestors in ways that didn't work correctly with replace-by-fee-rate. So I changed it in a later commit to GetFee.
I believe that will also work incorrectly in cases where fees are manually prioritized, as, IIUC, manual priorities are done via GetModifiedFee. But I haven't had time (or funding) to fix those issues yet. This code is just testing the idea out right now.
Good work!
reply
i love how short pretty powerful commits to bitcoin-core are
reply