This week’s newsletter announces the release of LND 0.8.2-beta, requests
help testing the latest C-Lightning release candidate, discusses
widespread support for basic multipath payments in LN, provides an update on
bech32 error detection reliability, summarizes updates to the proposed
OP_CHECKTEMPLATEVERIFY opcode, and links to a discussion about the impact
of eclipse attacks on LN channels. Also included are our regular
sections about notable changes to popular Bitcoin infrastructure
projects, changes to services and client software, and popular Bitcoin
Stack Exchange questions and answers.
● Help test C-Lightning 0.8.0 RC: the release candidate for the next version of C-Lightning switches the default network to mainnet instead of testnet (see Newsletter #75) and adds support for basic multipath payments (described below) among many other features and bug fixes.
● Review bech32 action plan: as described below, Pieter Wuille suggests restricting all bech32 addresses to 20 or 32 byte witness programs in order to prevent a loss of funds from transcription errors involving addresses ending with a
p. This rule already applies to v0 segwit addresses used for P2WPKH and P2WSH, so the change would simply be extending it to v1+ addresses currently reserved for future upgrades (such as the proposed taproot). This will require wallets and services that have already implemented bech32 sending support to upgrade their code, but it should be a tiny change; e.g. for the Python reference implementation it might look something like:
--- a/ref/python/segwit_addr.py +++ b/ref/python/segwit_addr.py @@ -110,7 +110,7 @@ def decode(hrp, addr): return (None, None) if data > 16: return (None, None) - if data == 0 and len(decoded) != 20 and len(decoded) != 32: + if len(decoded) != 20 and len(decoded) != 32: return (None, None) return (data, decoded)
If you have any questions or concerns about this suggested change, please reply to the mailing list post linked in the news section below.
● LN implementations add multipath payment support: after a significant amount of discussion and development, all three LN implementations tracked by Optech have now added basic support for multipath payments (merges: C-Lightning, Eclair, LND). A multipath payment consists of multiple LN payments routed over different paths which may all be claimed at the same time by the receiver. This provides a major upgrade to the usability of LN by allowing users to spend or receive funds in several of their channels in the same overall payment. This upgrade means that spenders in particular have much less need to worry about how much balance they have in any particular channel as they can send up to their full available balance across all their channels (subject to other LN limits).
● Analysis of bech32 error detection: Pieter Wuille sent an email to the Bitcoin-Dev mailing list following up on the bech32 malleability concerns described in previous newsletters (#72, #74, and #76) where it’s possible to add or remove any number of
qcharacters immediately before a
pcharacter at the end of a bech32 string. Wuille’s analysis shows that this is the only exception to bech32’s expected error detection properties and that “changing one constant in bech32 would resolve this issue.”
Wuille plans to amend BIP173 to describe the weakness, propose a change to limit existing bech32 address uses to either 20 byte or 32 byte witness program payloads, and define a modified version of bech32 with the alternative constant for non-Bitcoin uses and for a potential future where we want witness programs that are not 20 bytes or 32 bytes.
● Proposed changes to bip-ctv: Jeremy Rubin suggested several changes to the proposed soft fork addition of an
OP_CHECKTEMPLATEVERIFY(CTV) opcode. Most notably, the changes will remove the restriction that templates used by CTV cannot be derived from other data via Bitcoin Script. This update simplifies the modification to the Script language discussed in Newsletter #75. None of the updates modifies CTV’s behavior in a way that would significantly affect previously described usecases as far as we’re aware (but anyone who is aware of fundamental changes is encouraged to discuss them on-list).
● Discussion of eclipse attacks on LN nodes: Antoine Riard posted to the Lightning-Dev mailing list a description of two attacks possible against LN users if they are eclipse attacked and the attacker delays the relay of blocks. An eclipse attack occurs when all connections made by a full node or lightweight client are controlled by a single attacker, as might easily be the case when the attacker is an ISP or takes control of a user’s router. This gives the attacker full control over what data the node or client sends or receives. In the first attack, the eclipse attacker can send a revoked commitment transaction without the honest user learning about it in time to submit the corresponding penalty transaction, allowing the attacker to steal from the honest user. In the second attack, the attacker prevents the honest user from realizing that they need to broadcast the latest commitment transaction because one or more of its HTLCs is about to expire—this allows the attacker to steal the funds in the HTLC after it does expire.
Both Riard’s post and replies from Matt Corallo and ZmnSCPxj discuss past and present work to make full nodes and lightweight clients more resistant to eclipse attacks. Readers interested in learning more about eclipse attacks and their mitigations are also strongly encouraged to read the Bitcoin Core review club’s meeting notes and log for the past week, as they covered this topic in depth.
Changes to services and client software
In this monthly feature, we highlight interesting updates to Bitcoin wallets and services.
● Bitfinex supports LN deposits and withdrawals: In a recent blog post, Bitfinex has announced support on their exchange for Lightning Network. Users of Bitfinex can now both deposit and withdraw funds using LN.
● BitMEX Research launches an LN penalty transaction tracker: In an article posted by BitMEX Research, the open source ForkMonitor tool now lists Lightning penalty transactions. The tool also monitors chaintips across a variety of Bitcoin implementations and versions in order to detect forks.
● BitMEX bech32 send support: In a recent blog post, BitMEX has announced support on their exchange for sending to native bech32 addresses. The post also outlines plans to migrate from P2SH to P2SH-wrapped segwit addresses for their own wallet.
● Unchained Capital open sources Caravan, a multisig coordinator: With a blog post and demo video, Unchained Capital has open sourced their multisig coordinator named Caravan. Caravan is a stateless web application for creating and spending from a multisig address using a variety of external keystores.
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.
● What is the rationale for the Lightning network’s path length limit (20 hops)? Sergei Tikhomirov asks about BOLT4’s 20 hop limit and the comparison of Tor’s onion routing with LN’s Sphinx. Rene Pickhardt outlines the differences in protocols and motivations for the current limit: keeping TCP/IP packages small and assumptions that LN be a small-diameter network.
● Is there a way to allow use of unconfirmed RBF outputs in transaction building? G. Maxwell points out that Bitcoin Core treats outputs from transactions signaling opt-in Replace-by-Fee (RBF) the same way it treats outputs from transactions that don’t signal RBF support. Differences in the way an output is handled by Bitcoin Core depend on whether the transaction containing the output is confirmed or not, and whether the transaction was created by the user’s Bitcoin Core wallet or not.
● What is the max allowed depth for BIP32 derivation paths? Andrew Chow explains that, since BIP32 allocates one byte for the depth field, there are a maximum of 256 possible elements in the derivation path.
Notable code and documentation changes
● Bitcoin Core #12763 adds an RPC whitelist feature that allows you to restrict which RPCs a particular user can run. By default, any authenticated user can run any command, but the new configuration options
rpcwhitelistdefaultmay be used to configure which users can access which RPCs.
● LND #3697 sets the default minimum HTLC value to 1 millisatoshis (msat) for new channels, down from a previous default of 1,000 msat. The minimum HTLC value can’t be changed once a channel is opened, so this change allows channels using this setting to accept sub-satoshi payments. [Edit: a previous version of this paragraph incorrectly claimed the new minimum was 0 msat; the correct value is 1 msat.]
● LND #3702 extends the
closechannelRPC with a
delivery_addressparameter that can be used to request a mutual close of a channel send funds to the specified address. This won’t work if the user previously activated the upfront shutdown script feature described in last week’s newsletter.
● LND #3415 allows settling invoices via multipath payments, adding the final necessary code for basic multipath payment support in LND (see the description of multipath payments in the news section above).
● BOLTs #643 adds basic multipath payment support as described in the news section above. This achieves one of the major goals set during the Lightning Specification 1.1 meeting a year ago to help significantly improve LN wallet UX.
Holiday publication schedule
We will not be publishing newsletters on December 25th or January 1st. Instead, we’ll publish our second annual year-in-review special report on Saturday, December 28th. Regular newsletter publication will resume on Wednesday, January 8th.