Because of an incorrect test, it wasn't discovered that the scriptSig
field was being set on the unsigned TX inputs for a nested SegWit input.
This commit fixes the bug and also refactors the test so it would have
caught this specific bug.
To fix a bug where specifying multiple UTXOs that are by themselves
large enough to satisfy the output amount would lead to the rest of them
being added to fees, we need to provide the transaction author with a
constant list of UTXOs. If we didn't, the author would only consider one
input and calculate the change based on that alone. But since we'd add
all inputs to the PSBT, the rest of the amounts would go to fees.
To make it easy to show the user what change output was created (if any)
during the funding process, we return its index (or -1 if no change
output was created).
We would previously request spend notifications for all transaction
outputs, leading to irrelevant transactions being found in the wallet's
transaction store.
In this commit, we fix a 3 year old bug in dust calculation. Before this
commit, the target fee of the transaction to be crafted would be used to
determine dust. If the target fee is very high, then this would cause
over all higher fees, as we'd base that dust computation off of that fee
rate, rather than the min relay fee.
To fix this, we always use the min relay fee at all times when computing
dust.
Not setting this would result in a non-sensible unix timestamp
(2288912640) being exposed when the wallet hasn't synced any blocks,
like in the case when it's waiting for the backend to sync.
This commit allows for the ability to lease an output to a particular ID
for a limited amount of time, ensuring that no other processes can use
said output for their coin selection needs. An output can either be
unlocked manually, or lazily whenever required.
Separate out the logic we use to write labels so that it can be used
in dropwtxmgr to re-add labels if we want to keep them when we drop the
rest of the db.
There's no need to retrieve the full block as we're only interesting in
retrieve its corresponding height, which can be done with
GetBlockHeaderVerbose.
Code is duplicated across these two functions, so update
rangeBlockTransactions to use minedTxDetails. When iterating through
a block, we do have the block metadata already (which is looked up by
minedTxDetails). This change can be further optimized to split
minedTxDetails into minedTxDetails and minedTxDetailsWithoutBlock to
reduce this redundancy.
