/ home / newsletters /
Bitcoin Optech Newsletter #210
This week’s newsletter describes a proposed BIP for creating signed messages for non-legacy addresses and summarizes a discussion about provably burning small amounts of bitcoin for denial of service protection. Also included are our regular sections with popular questions and answers from the Bitcoin Stack Exchange, announcements of new releases and releases candidates, and summaries of notable changes to popular Bitcoin infrastructure software.
News
-
● Multiformat single-sig message signing: Bitcoin Core and many other wallets have long included support for signing and verifying arbitrary messages when the key used to sign them corresponds to a P2PKH address. Bitcoin Core doesn’t support signing or verifying arbitrary messages for any other address type, including addresses covering single-sig P2SH-P2WPKH, native P2WPKH, and P2TR outputs. A previous proposal, BIP322, to provide fully generic message signing that could work with any script has not yet been merged into Bitcoin Core or added to any other popular wallet of which we’re aware.
This week, Ali Sherief proposed that the same message signing algorithm used for P2WPKH also be used for other output types. For verification, programs should infer how to derive the key (if necessary) and verify the signature using the address type. E.g., when provided a bech32 address with a 20 byte data element, assume it’s for a P2WPKH output.
Developer Peter Gray noted that ColdCard wallets already create signatures in this way and developer Craig Raw said Sparrow Wallet that wallet is able to validate them in addition to also following the BIP137 validation rules and a slightly different set of rules implemented in Electrum.
Sherief is planning to write a BIP specifying the behavior.
-
● Proof of micro-burn: several developers discussed use cases and designs of onchain transactions that destroy bitcoins (“burn” bitcoins) in small increments as a proof of resource consumption. To extend an example use case by Ruben Somsen from the thread, the idea would be to allow 100 users to each attach to their emails a proof that $1 of bitcoins had been burned, providing the type of anti-spam protection originally envisioned as a benefit of hashcash.
Several solutions were discussed using merkle trees, although one respondent suggested that the small amounts involved suggest that having participants trust (or partially trust) a centralized third party may be a reasonable way to avoid unnecessary complexity.
Selected Q&A from Bitcoin Stack Exchange
Bitcoin Stack Exchange is one of the first places Optech contributors look for answers to their questions—or when we have a few spare moments to help curious or confused users. In this monthly feature, we highlight some of the top-voted questions and answers posted since our last update.
-
● Why do invalid signatures in
OP_CHECKSIGADD
not push to the stack? Chris Stewart asks why “if an invalid signature is found, the interpreter fails execution rather than continuing”. Pieter Wuille explains that this behavior, defined in BIP340-342, is designed to support batch validation of schnorr signatures in the future. Andrew Chow gives an additional reason for the behavior, noting that certain malleability concerns are also mitigated by this approach. -
● What are packages in Bitcoin Core and what is their use case? Antoine Poinsot explains packages (a grouping of related transactions), their relation to package relay, and a recent package relay BIP proposal.
-
● How much blockspace would it take to spend the complete UTXO set? Murch explores a hypothetical scenario of consolidating all existing UTXOs. He provides blockspace calculations for each output type and concludes the process would take about 11,500 blocks.
-
● Does an uneconomical output need to be kept in the UTXO set? Stickies-v notes that while provably unspendable UTXOs including
OP_RETURN
or scripts larger than the max script size are removed from the UTXO set, removing uneconomical outputs could cause issues, including a hard fork as Pieter Wuille points out, if those outputs are spent. -
● Is there code in libsecp256k1 that should be moved to the Bitcoin Core codebase? Similar to other efforts to modularize areas of the Bitcoin Core codebase like libbitcoinkernel or process separation, Pieter Wuille notes a clear area of responsibility of the libsecp256k1 project: everything that involves operations on private or public keys.
-
● Mining stale low-difficulty blocks as a DoS attack Andrew Chow explains that assumevalid and more recently
nMinimumChainWork
help filter out low-difficulty chain attacks.
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.
-
● BTCPay Server 1.6.3 adds new features, improvements, and bug fixes to this popular self-hosted payment processor.
-
● LDK 0.0.110 adds a variety of new features (many covered in previous newsletters) to this library for building LN-enabled applications.
Notable code and documentation changes
Notable changes this week in Bitcoin Core, Core Lightning, Eclair, LDK, LND, libsecp256k1, Hardware Wallet Interface (HWI), Rust Bitcoin, BTCPay Server, BDK, Bitcoin Improvement Proposals (BIPs), and Lightning BOLTs.
-
● Bitcoin Core #25351 ensures that after an import of addresses, keys, or descriptors to a wallet, the subsequent rescan will not only scan the blockchain but also evaluate whether transactions in the mempool are relevant to the wallet.
-
● Core Lightning #5370 reimplements the
commando
plugin and makes it a built-in part of CLN. Commando allows a node to receive commands from authorized peers using LN messages. Peers are authorized using runes, which is a custom CLN protocol based on a simplified version of macaroons. Although Commando is now built into CLN, it’s only operable if a user creates rune authentication tokens. For additional information, see CLN’s manual pages for commando and commando-rune. -
● BOLTs #1001 recommends that nodes who advertise a change to their payment forwarding policies continue accepting payments received using the old policies for about 10 minutes. This prevents payments from failing just because the sender hasn’t heard about a recent policy update. See Newsletter #169 for the example of an implementation adopting a rule like this.