Closed kristapsk closed 1 month ago
laanwj commented on Apr 12
Correct, version 28.0 (development on current master now that 27.0 is branched off) is extremely likely not going to have legacy wallet support anymore. We'll have to find some solution for this.
https://github.com/JoinMarket-Org/joinmarket-clientserver/issues/1571#issuecomment-2051080385
laanwj commented on Apr 12
Correct, version 28.0 (development on current master now that 27.0 is branched off) is extremely likely not going to have legacy wallet support anymore. We'll have to find some solution for this.
Ahh, yes, right, totally forgot about this (really busy with other stuff recently).
As there is not much development activity recently, guess we will need to just implement hackish descriptor wallet support I proposed in #1064.
There is also another problem, that even with my previously proposed PR, tests would still not be compatible with v28 and descriptor wallets in general, as currently we have single Bitcoin Core wallet which are used both for mining new coins and then spending them and as a watch-only wallet, but descriptor wallets does not allow importing addresses into wallets with enabled private keys.
The test case failure in this issue is unrelated to legacy-vs-descriptor wallets. This issue is due to the change in Bitcoin Core 28.0 of the default value for the mempoolfullrbf
config option from 0 to 1. You can make the test_check_valid_candidate_not_replaceable
test case pass by adding this line to the bitcoin.conf
used for the tests:
mempoolfullrbf=0
In the long run, though, the test case is arguably ill-conceived (or at best obsolete). It has never been safe to assume that an unconfirmed transaction could not be replaced merely because it did not "opt in" to RBF "eligibility."
Thanks @whitslack for the hint!
See https://github.com/kristapsk/joinmarket-clientserver/actions/runs/11199073884/job/31131015181 and https://github.com/kristapsk/joinmarket-clientserver/actions/runs/11199169029/job/31131227626.