Skip to content

Logbook July 2026

Paolo Veronelli edited this page Jul 24, 2026 · 2 revisions

Week of Jul 6th — The project goes public: cardano-aid is renamed to cardano-keri across the Haskell namespace, the Aiken package, and the value-write signing domain, the repository is opened with meeting material kept out of the tree, and the docs site starts deploying to GitHub Pages. The week then turns technical — the identity specs (detached-signature authorization envelope, key-state UTxO, permissionless pre-rotation) are frozen and the org-standard CI shape is restored so tests actually run. The central question, whether BLAKE3 genesis verification can fit a Plutus transaction, is chased from a flat "does not fit" verdict through four optimization rounds to a lane-packed implementation where the whole single-chunk domain fits the mainnet budget.

Tuesday 7th

Public release and docs deploy (PV)

Identity spec groundwork (PV)

Wednesday 8th

Primers and model vetting (PV)

CI hardening (PV)

  • Merged ci: org-standard CI shape #66 — CI was a false green that built artifacts but never ran their tests; now runs unit tests, adds a dev-shell cabal gate and cachix, with a justfile mirroring CI across both planes (closes #56, #57).

Thursday 9th

Identity model design (PV)

Architecture and economics docs (PV)

BLAKE3-in-Plutus spike (PV)

Friday 10th

BLAKE3 optimization rounds (PV)

Saturday 11th

Cage security fix and genesis decision (PV)

Week of Jul 13th — With BLAKE3 proven to fit, the focus shifts to freezing the checkpoint's on-chain shape and starting to build it. Midweek settles the checkpoint contention model (a sovereign per-AID UTxO over an MPFS trie) and freezes the weighted CheckpointDatumV1 wire contract with byte-for-byte Haskell/Aiken parity. The week then turns the ratified convict/freeze divergence-enforcement policy into real spend paths proven against the keripy reference, corrects the witness-rotation model to KERI's actual incoming-set rule, and lands the first registration-path validator.

Wednesday 15th

Checkpoint contention model (PV)

Thursday 16th

Freeze CheckpointDatumV1 (PV)

  • Merged design(identity): freeze weighted CheckpointDatumV1 and CBOR boundary #105 — freezes the schema-support layer #24 will consume: the e-native checkpoint (standard Blake3 AIDs register as-is), KERI-faithful pre-rotation with an explicit next pair and dual-threshold advance, and byte + verdict parity across Haskell and Aiken (resolves #68, absorbs #77/#79/#81).

Friday 17th

Divergence-enforcement docs (PV)

  • Merged docs(106): convict/freeze divergence enforcement #107 — documents the ratified enforcement policy across the identity model, trust docs, roadmap, and deck: the three-mode table (fork → convict/burn, behind → freeze, ahead → preventively mitigated).

Saturday 18th

Divergence-enforcement spend paths (PV)

  • Merged feat(identity): convict/freeze divergence-enforcement spend paths #111 — the validator-free schema-support layer for on-chain enforcement of KERI↔Cardano divergence, proven against the keripy reference and mirrored byte-for-byte and verdict-for-verdict across Haskell and Aiken; convict evidence now requires witness receipts (anti-fork) (resolves #106).

Presentation and witness-rotation correction (PV)

Sunday 19th

Registration path (PV)

Week of Jul 20th — The enforcement lifecycle comes together and is then torn back to its ratified form. Monday wires freeze, conviction, and fixed-bond enforcement onto the advance path; the rest of the week reverses conviction from permanent identity retirement to penalty-plus-record and redesigns registration and freeze as permissionless bonded state machines. In parallel the whole M1 lifecycle is proved in Lean — first 17 theorems, then 21 under the new "burn axiom" — before a compliance audit gives the pipeline a GO.

Monday 20th

Advance path and enforcement wiring (PV)

Tuesday 21st

Bonded freeze and the Lean model (PV)

Wednesday 22nd

Burn axiom, audit, and permissionless registration (PV)