Unintuitive and undocumented `full_scan` behavior

[andreasgriffin]

Book of BDK has a overview of the expected sync behavior

However full_scan seems to completely ignore the revealed addresses and goes by the gap only. This means that if there is a large gap it will not discover any transactions even though the addresses are already revealed. A sync will discover those transactions, such that for a "full scan" one has to do full_scan and sync .

Thanks to @design-rrr for finding this behavior.

• Reproducible in bdkpython 2.0.0

Why not do sync most of the time?

• onchain wallets can have multiple instances running, that reveal and use scripts. That means that the wallet always has to sync at least the gap limit ahead to discover txs that other instances used.

What to do about it?

• Was this behavior intended?
  • But then to do a full scan, one has to hit the server twice... --> inefficient

@thunderbiscuit @ValuedMammal

[ValuedMammal]

To avoid further confusion I want to propose that we eliminate full-scan and consolidate the sync and full-scan logic into a single sync workflow that keeps the existing sync behavior while making use of the stop-gap functionality.

[oleonardolima]

I've moved this issue to bitcoindevkit/bdk, as it applies to bdk_electrum/esplora and not the rust-electrum-client.