Address gap limit?

[evanlinjin]

Describe the situation

An undisclosed user of ours ran into an issue where creating multiple unbroadcasted transactions resulted in incrementing the revealed index for the internal keychain too rapidly. Eventually, the gap of unused addresses surpassed the stop_gap set for the spk-based chain source, thus some transactions were not caught by the syncing process.

This is caused by Wallet::create_tx which marks change addresses as used.

bdk/crates/wallet/src/wallet/mod.rs

Line 1460 in 775e4ae

self.indexed_graph.index.mark_used(change_keychain, index);

The correct use of the current API would be to call Wallet::cancel_tx. However, I would argue that this is unwieldy and not immediately apparent.

bdk/crates/wallet/src/wallet/mod.rs

Line 1963 in 775e4ae

pub fn cancel_tx(&mut self, tx: &Transaction) {

Also, recently another user mentioned trying to implement some sort of address-recycling logic on Discord @ #bdk-users.

What to do

Not sure, let's discuss.

[notmandatory]

Is this fixed by #1579 ?

[LLFourn]

This is an issue about a new feature that would stop revealing new addresses passed the stop gap. It's related to #1579 as it would have softened the problem for the user who doesn't called cancel_tx but it is not fixed by resolving that.

[notmandatory]

If this a new feature request can we move it out of the 1.0 beta milestone?