In case anyone is interested, a few months ago I resurrected my old ln-symmetry branch of CLN with various improvements and examples, thanks to LLM doing the heavy lifting for revival.
Aside from APO implementation, I also vibed a CSFS+TH+IK variant, though it cannot be tested on signet directly until OP_TEMPLATEHASH is activated. It's a drop-in replacement for CTV, if anyone cares to adapt it.
Greg
On Tuesday, April 14, 2026 at 5:18:24 AM UTC-4 aaron.recompile wrote:
Hi List,
I ran a three-round Eltoo-style state chain on Bitcoin Inquisition 29.2
signet using APO (BIP 118) for state updates and CTV for settlement.
Six transactions, all confirmed on-chain.
An independent Python implementation of the BIP 118 sighash (Msg118 /
Ext118) was cross-validated against Inquisition's C++ consensus engine.
If the Python digest disagrees with Core's, the transaction is rejected.
It wasn't.
Construction
------------
Each state UTXO is a three-leaf Taproot tree:
leaf 1 -- ctv_uhpo: <template_hash> OP_CHECKTEMPLATEVERIFY
leaf 2 -- apo_update: <0x01||xonly> OP_CHECKSIG (BIP 118)
leaf 3 -- csv_escape: <sequence> OP_CHECKSEQUENCEVERIFY OP_DROP
<pubkey> OP_CHECKSIG
CTV locks each state's payout distribution
as a template hash. APO rebinding advances state. When an APO update
spends state vN, vN's CTV leaf
loses its only viable input and can never settle. Only the latest state
can produce a valid settlement. This is the Eltoo invariant, executed
end-to-end with current tooling.
Confirmed txids (Inquisition signet)
-------------------------------------
Fund → State v1:
386dbb6aa23fcc35a69d34e3c0f760b185482467abc936196d3def19d54a9c41
APO update v1 → v2:
096e31ccbd8f5460b2730ec4f757ee1b01acf9dde3e3b8cb55fbb534fe195601
APO update v2 → v3:
091309b73e299436ff12fceda7e54e28c6f1817fdf3405bca3343ad15198d7f6
CTV settlement: 13957f49d01aa21ed2aa28df7aa78f357d51a2a2fbfb434af05fdc75ad0ff9b7
Rebind A (signed):
03c0577c1d47da32804d098187644d0eee18b448aded2f427cd02193c070f3a4
Rebind B (witness copied):
46091190c74d8fd4b39be67a2e945a19b021850e7f8d9e378f5eb11722ae1a43
Rebind A and B share an identical witness stack — same 65-byte signature
(sighash flag 0x41), same leaf script, same control block — spending two
different UTXOs without re-signing. Outputs remain committed: changing
the output address in Rebind B causes script verification failure.
The last two transactions are a standalone rebinding demonstration,
not part of the state chain above.
Links
------
BIP 118 signing primitive and rebinding proof (Delving Bitcoin):
https://delvingbitcoin.org/t/bip-118-signing-from-scratch-on-chain-rebinding-proof/2411
Eltoo state chain construction (Delving Bitcoin):
https://delvingbitcoin.org/t/eltoo-state-chain-on-signet-three-rounds-six-transactions-apo-ctv/2413
Source code (btcaaron, examples/braidpool/):
https://github.com/aaron-recompile/btcaaron
Happy to discuss the construction or the CTV+CSFS gap.
Aaron Zhang
You received this message because you are subscribed to the Google Groups "Bitcoin Development Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bitcoindev+unsubscribe@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/bitcoindev/494c0600-2daa-4500-8e49-bf36efbf2625n%40googlegroups.com.