This week’s newsletter announces a new domain-specific language for experimenting with contract protocols, summarizes a discussion about modifying BIP editor responsibilities, and describes proposals to reset and modify testnet. Also included are our regular sections with the summary of a Bitcoin Core PR Review Club meeting, announcements of new releases and release candidates, and descriptions of notable changes to popular Bitcoin infrastructure software.

News

  • DSL for experimenting with contracts: Kulpreet Singh posted to Delving Bitcoin about a domain-specific language (DSL) he’s working on for Bitcoin. The language makes it easy to specify the operations that should be performed as part of a contract protocol. This can make it easy to quickly execute the contract in a testing environment to ensure it behaves as expected, allowing fast iteration on new ideas for contracts and providing a baseline against which later full-fledged software can be developed.

    Robin Linus replied with a link to a somewhat similar project to allow a higher-level language to describe a contract protocol that will be compiled down to the necessary operations and low-level code to execute that protocol. This work is being done as part of enhancing BitVM.

  • Updating BIP2: Tim Ruffing posted to the Bitcoin-Dev mailing list about updating BIP2, which specifies the current process for adding new BIPs and updating existing BIPs. Several problems with the current process mentioned by Ruffing and others included:

    • Editorial evaluation and discretion: how much effort should BIP editors be required to expend on ensuring new BIPs are high quality and focused on Bitcoin? Separately, how much discretion should they have in being able to reject new BIPs? Ruffing and several others mentioned that they’d prefer minimizing editorial requirements and privileges, perhaps depending on BIP editors only to prevent systemic abuse (e.g. mass spamming). Of course, BIP editors—like any other community members—would be able to voluntarily suggest improvements to any BIP proposals they found interesting.

    • Licensing: some allowed licenses for BIPs are designed for software and may not make sense for documentation.

    • Comments: as a change from BIP1, BIP2 attempted to provide a place for community feedback on each BIP. This has not been widely used and the results have been controversial.

    The idea of updating BIP2 was still being discussed at the time of writing.

    In a separate but related discussion, the nomination and advocacy for new BIP editors mentioned in last week’s newsletter has been extended to the UTC end of day on Friday, April 19th. It is hoped that the new editors will receive merge access by the end of the day on the following Monday.

  • Discussion about resetting and modifying testnet: Jameson Lopp posted to the Bitcoin-Dev mailing list about problems with the current public Bitcoin testnet (testnet3) and suggested restarting it, potentially with a different set of special-case consensus rules.

    Previous versions of testnet had to be restarted when some people began assigning economic value to testnet coins, resulting in them becoming hard to acquire for free for people who wanted to perform actual testing. Lopp provided evidence of that happening again and also described the well-known problem of block flooding due to exploitation of testnet’s custom difficulty adjustment algorithm. Multiple people discussed potential changes to testnet to address that and other problems, although at least one respondent preferred allowing problems to continue as it made for interesting testing.

Bitcoin Core PR Review Club

In this monthly section, we summarize a recent Bitcoin Core PR Review Club meeting, highlighting some of the important questions and answers. Click on a question below to see a summary of the answer from the meeting.

Implement 64 bit arithmetic op codes in the Script interpreter is a PR by Chris Stewart (GitHub Christewart) that introduces new opcodes allowing users to perform arithmetic operations on larger (64-bit) operands in Bitcoin Script than is currently allowed (32-bit).

This change, in combination with some existing soft fork proposals like OP_TLUV that enable transaction introspection, would allow users to build scripting logic based on a transaction’s satoshi-denominated output values, which can easily overflow a 32-bit integer.

Discussion on the approach, such as whether to upgrade existing opcodes or introduce new ones (e.g., OP_ADD64), is still ongoing.

For more information, see the (WIP) BIP, and the discussion on the Delving Bitcoin forum.

  • What does the CScriptNum nMaxNumSize parameter do?

    It represents the maximum size (in bytes) of the CScriptNum stack element being evaluated. By default, it is set to 4 bytes. 

  • What 2 opcodes accept 5-byte numeric inputs?

    OP_CHECKSEQUENCEVERIFY and OP_CHECKLOCKTIMEVERIFY use signed integers to represent time stamps. Using 4 bytes, that would put the upper range of allowed dates in 2038. For that reason, a carve-out was made for these 2 time-based opcodes to accept 5-byte inputs. This is documented in the code

  • Why was the fRequireMinimal flag introduced to CScriptNum?

    CScriptNum has a variable length encoding. As described in BIP62 (rule 4), this introduces opportunity for malleability. For example, zero can be encoded as OP_0, 0x00, 0x0000, … Bitcoin Core #5065 fixed this in standard transactions by requiring a minimal representation for data pushes and stack elements that represent numbers. 

  • Is the implementation in this PR malleability safe? Why?

    The current implementation requires a fixed-length 8-byte representation for the operands of a 64-bit opcode, making it safe against zero-padding malleability. The rationale is to simplify implementation logic, at the cost of increased blockspace usage. The author has also explored using a CScriptNum variable encoding in a different branch

Releases and release candidates

New releases and release candidates for popular Bitcoin infrastructure projects. Please consider upgrading to new releases or helping to test release candidates.

  • HWI 3.0.0 is a release of the next version of this package providing a common interface to multiple different hardware signing devices. The only significant change in this release is that emulated hardware wallets will no longer be automatically detected; see the description of HWI #729 below for details.

  • Core Lightning 24.02.2 is a maintenance release that fixes “a small incompatibility” between Core Lightning’s and LDK’s implementation of a particular part of the LN gossip protocol.

  • Bitcoin Core 27.0rc1 is a release candidate for the next major version of the network’s predominant full node implementation. Testers are encouraged to review the list of suggested testing topics.

Notable code and documentation changes

Notable recent changes in Bitcoin Core, Core Lightning, Eclair, LDK, LND, libsecp256k1, Hardware Wallet Interface (HWI), Rust Bitcoin, BTCPay Server, BDK, Bitcoin Improvement Proposals (BIPs), Lightning BOLTs, Bitcoin Inquisition, and BINANAs.

Note: the commits to Bitcoin Core mentioned below apply to its master development branch and so those changes will likely not be released until about six months after the release of the upcoming version 27.

  • Bitcoin Core #29648 removes libconsensus after it was previously deprecated (see Newsletter #288). Libconsensus was an attempt to make Bitcoin Core’s consensus logic usable in other software. However, the library hasn’t seen any significant adoption and it has become a burden on maintenance of Bitcoin Core.

  • Bitcoin Core #29130 adds two new RPCs. The first will generate a descriptor for a user based on the settings they want and then add that descriptor to their wallet. For example, the following command will add support for taproot to an old wallet created without that support:

    bitcoin-cli --rpcwallet=mywallet createwalletdescriptor bech32m
    

    The second RPC is gethdkeys (get HD keys), which will return each xpub used by the wallet and (optionally) also each xpriv. When a wallet contains multiple xpubs, the particular one to use can be indicated when calling createwalletdescriptor.

  • LND #8159 and #8160 add experimental (disabled by default) support for sending payments to blinded routes. A follow-up PR is expected to add complete error handling for failed blinded payments.

  • LND #8515 updates multiple RPCs to accept the name of the coin selection strategy to be used. See Newsletter #292 for the previous improvements to the flexibility of LND’s coin selection that this PR builds upon.

  • LND #6703 and #6934 add support for inbound routing fees. A node can already advertise the cost it will charge to forward a payment through a particular outbound channel. For example, Carol might advertise that she will only forward payments to her channel peer Dan if the payments offer 0.1% of their value to her. If that lowers the average number of satoshis (sats) per minute that Carol forwards to Dan below the average amount he forwards to her, eventually all of the channel balance will end up on Carol’s side, preventing Dan from being able to forward more payments to her, reducing both of their revenue potentials. To prevent that, Carol might lower her outbound forwarding fee to Dan to 0.05%. Similarly, if Carol’s lower outbound forwarding fee to Dan results in her forwarding more sats per minute to him than he forwards to her, all of the balance might end up on his side of the channel, also preventing additional forwarding and revenue earning; in that case, Carol can raise her outbound fees.

    However, outbound fees only apply to outbound channels. Carol is offering to charge the same fee regardless of what channel she receives the payment over; for example, she charges the same rate whether she receives the payment from either of her channel peers Alice or Bob:

    Alice -> Carol -> Dan
    Bob -> Carol -> Dan
    

    This makes sense since the base LN protocol doesn’t pay Carol for receiving a forwarding request from Alice or Bob. Alice and Bob can set outbound fees for their channels to Carol, and it’s up to them to set fees that help keep the channels liquid. Similarly, Carol can adjust her fees for outbound payments to Alice and Bob (e.g. Dan -> Carol -> Bob) to help manage liquidity.

    However, Carol may want more control over policies that affect her. For example, if Alice’s node is poorly managed, she might frequently forward payments to Carol without many people later wanting to forward payments from Carol to Alice. That would eventually end with all the funds in their channel ending up on Carol’s side, preventing further payments in that direction. Before this PR, there was nothing Carol could do about that, except close her channel with Alice before it wasted too much of the value of Carol’s capital.

    With this PR, Carol can now also charge an inbound forwarding fee that is specific to each channel. For example, she might charge a high fee for payments arriving inbound from Alice’s problematic node but a lower fee for payments arriving inbound from Bob’s highly liquid node. Initial inbound fees are expected to always be negative to make them backward compatible with older nodes that don’t understand inbound fees; for example, Carol might give a 10% fee discount on payments forwarded by Bob and a 0% discount on payments forwarded by Alice.

    The fees are assessed simultaneously with the outbound fees. For example, when Alice offers a payment to Carol for forwarding to Dan, Carol calculates the original dan_outbound fee, calculates the new alice_inbound fee, and ensures the forwarded payment offers her at least the sum of both. Otherwise, she rejects the HTLC. Since the initial inbound fees are always expected to be negative, Carol won’t reject any payments that pay sufficient outbound fees, but any node that’s now aware of inbound fees may be able to receive a discount.

    Inbound routing fees were first proposed about three years ago, discussed on the Lightning-Dev mailing list about two years ago, and documented in draft BLIPs #18 also about two years ago. Since its initial proposal, several maintainers of LN implementations besides LND have opposed it. Some have opposed it on principle; others have opposed its design as being overly specific to LND rather than a local and generic upgrade that can immediately use positive inbound forwarding fees and doesn’t require global advertisement of additional fee details for each channel. An alternative approach is proposed in draft BLIPs #22. We’re only aware of one maintainer of a non-LND implementation indicating that they’ll adopt LND’s method—and only in cases where negative inbound forwarding fees are offered, as that’s “free money for our users.”

  • Rust Bitcoin #2652 changes what public key is returned by the API when signing a taproot input as part of processing a PSBT. Previously, the API returned the public key for the signing private key. However, the PR notes that “it is common to think of the internal key as being the one that signs even though this is not technically true. We also have the internal key in the PSBT.” With the merge of this PR, the now API returns the internal key.

  • HWI #729 stops automatically enumerating and using device emulators. Emulators are mainly used by developers of HWI and hardware wallets, but attempting to automatically detect them may cause issues for regular users. Developers who do want to work with emulators now need to pass an additional --emulators option.