The fee estimation data format changed from 0.14.x to 0.15.0, so we should no longer read the old data. H/T @jnewbery, @morcos
Pending testing.
Is this just a cleanup or does this result in crashes/misestimations?
968 | - for (unsigned int i = 0; i < tempBuckets.size(); i++) { 969 | - tempMap[tempBuckets[i]] = i; 970 | - } 971 | - } 972 | - else { // nVersionThatWrote >= 149900 973 | + if (nVersionThatWrote > 149900) {
This will always evaluate to false, as nVersionThatWrote is currently ==149900
I don't think so - on 0.15.0 it should be 150000 on master it should be 159900
Mixed that up as nVersionRequired, sorry.
@laanwj This is meant to just be cleanup. The code to process the old file was left in under the thought we might provide a transition, but then we never did. Which reminds me, we should mention in the release notes that upgrading starts your estimates over again.
There are also similar cleanups to be had in TxConfirmStats::Read.
@Xekyo care to respond to @TheBlueMatt's comment? There are a couple of if statements in TxConfirmStats::Read that are now redundant:https://github.com/bitcoin/bitcoin/blob/67879b7c443144e479c763f64abb9c1ed7e44eaa/src/policy/fees.cpp#L415 and https://github.com/bitcoin/bitcoin/blob/67879b7c443144e479c763f64abb9c1ed7e44eaa/src/policy/fees.cpp#L444
This PR should also clean up that code.
Ping @Xekyo , if you include @TheBlueMatt's nits this is ready for merge.
utACK 3a3a9f9 I'm going to merge this. Removing this redundant code is clearly an improvement, and the other things suggested can be done later (added "up for grabs" tag).
utACK cdd6bbf.