Release v0.7.0

By Alekos Filini on 5/18/2021 - Tags: Rust, Release

A new release of BDK is out: v0.7.0 (opens new window) improved APIs, a more robust signing process and various bugfixes and improvements.

You can find the full v0.7.0 changelog (opens new window) on GitHub.

# What's new in v0.7.0

Below are some highlights of the new release:

# New Signing API

The Wallet::sign() method has been updated to take a mutable reference to a PSBT rather than consuming it entirely. This makes it easier to sign the same PSBT multiple times with different signers:

let mut psbt = ...;

let alice_finalized = wallet_alice.sign(&mut psbt, SignOptions::default())?;
let bob_finalized = wallet_bob.sign(&mut psbt, SignOptions::default())?;

Under the hood, even more has changed around the way we make signatures: starting from this release, our signer will require by default that SegWit PSBTs also provide the non_witness_utxo to mitigate the fee bruning "SegWit bug (opens new window)".

For this reason, the second argument of Wallet::sign(), which was initially used to optionally provide a different "current height" that the wallet would consider when trying to finalize the transaction, has been replaced with a more generic "options" argument of type SignOptions (opens new window). This allows setting a different assumed block height and also opting out of requiring non_witness_utxo by enabling the trust_witness_utxo flag.

As a result of these new more strict requirements, BDK will also, by default, fill-in the non_witness_utxo field when creating new PSBTs. The TxBuilder::force_non_witness_utxo() option has been removed and a new one called TxBuilder::only_witness_utxo() has been added, to allow users to opt-out of this new default behavior.

// Make a transaction that only contains the `witness_utxo`
let (psbt, details) = {
    let mut builder = wallet.build_tx();
        .add_recipient(send_to.script_pubkey(), 50_000)
// Sign it by explicitly trusting just the `witness_utxo`
let finalized = wallet.sign(&mut psbt, SignOptions { trust_witness_utxo: true, ..Default::default() })?;

# Support Timelocks in the policy Module

A less-known part of BDK is its "policy" module: its goal is to take a descriptor and encode the spending policy represented by that descriptor in a more "user-friendly" format. On top of that, the module tries to provide a summary for what a user's descriptor can contribute to a transaction. For instance, given a 2-of-2 multisig policy, a descriptor that contains only the two public keys can't contribute anything, while a descriptor that has one or both private keys can, respectively, contribute to and satisfy the policy by making signatures.

In release v0.5.0 we added support for computing which parts of a policy are already satisfied by a given PSBT. This, combined with the contribution part, allow users to get a complete picture of what's already present and what's missing to fully satisfy a descriptor.

In this release we are starting to take timelocks into consideration when computing the satisfaction component of a policy: this means that we can consider timelocks that are already expired as fully satisfied and also exclude policy branches that require specific nLockTime or nSequence values, if those aren't correctly set in the transaction.

Ultimately with those changes we are able to give our users a more complete picture of the completion stage of a PSBT, which also takes into account the expiration of timelocks.

# Contributors

A huge thanks to everybody who contributed to this new release with suggestions, pull requests and bug reports.

Since the v0.6.0 release around a month ago, we've had 39 new commits made by 6 different contributors for a total of 698 additions and 309 deletions. Here's the full diff (opens new window).

A special thanks to the new contributor for this release: