[bodymindarts]

Summary

• Bump es-entity 0.19.2 → 0.20.0, job 0.18.3 → 0.18.5, obix 0.14.3 → 0.14.4 together (they share es-entity types). errlanes follows es-entity. Both lockfiles regenerated by cargo; fuzz/Cargo.lock was stale on main (es-entity 0.17, cala 0.32.2-dev) and moves forward too.
• The eight entities (Account, AccountSet, Entry, Journal, Transaction, TxTemplate, VelocityControl, VelocityLimit) hydrate via the generated X::hydrator() / build(events).
• derive_builder is replaced by bon = "3" everywhere and removed from every manifest. cargo tree -i derive_builder matches nothing for cala-ledger, cala-ledger-core-types, cala-cel-interpreter, cala-perf, cala-ledger-example-rust and the fuzz crate. cala-ledger-core-types never used derive_builder and simply drops it.
• Identifiers removed between es-entity 0.19.2 and 0.20.0 (errlanes Carrier/Absorb/AbsorbFailBy/Repr/kind/LaneRef, narrow*, fatal(..) lift cases, UninitializedFieldError): only UninitializedFieldError-style hydration was used in cala. job 0.18.5 kept JobSvcConfig's shape; obix 0.14.4's MailboxConfig::build() is infallible (outbox/publisher.rs drops .expect).

API changes (breaking)

Infallible build() (bon typestate — a missing required field is now a compile error):
NewAccount, NewAccountSet, NewJournal, NewTxTemplate, NewVelocityEnforcement, NewEntry, NewTransaction, AccountUpdate, AccountSetUpdate, JournalUpdate.

Fallible build() with a new error type (validation unchanged, now a #[bon] fallible constructor; messages no longer go through to_string()):

builder	was	now
NewTxTemplateEntry, NewTxTemplateTransaction, NewVelocityControl, NewVelocityLimit, NewPartitionKey, NewLimit, NewBalanceLimit	XBuilderError (UninitializedField / ValidationError(String))	cel_interpreter::CelParseRejection
NewParamDefinition	NewParamDefinitionBuilderError	InvalidParamDefinition (Parse, Evaluation delegate to the CEL rejections; DefaultTypeMismatch { specified, source }, code CALA_PARAM_DEFAULT_TYPE_MISMATCH)
CalaLedgerConfig	CalaLedgerConfigBuilderError	InvalidCalaLedgerConfig (NoConnection, BothConnections; codes CALA_LEDGER_CONFIG_*)

Update types: AccountUpdate, AccountSetUpdate, JournalUpdate are now plain structs with #[derive(bon::Builder, Default)]; AccountUpdateValues, AccountSetUpdateValues, JournalUpdateValues are removed. X::default() + &mut setters + .build()? becomes X::builder().<setters>.build(). update() still takes impl Into<XUpdate>, and X::default() is still an empty update.

Setters: names, into conversion and Option setters taking T are kept; every optional member also gets maybe_x(Option<_>). Setters consume the builder (owned) instead of &mut self, so statement-style b.x(..); on a let mut builder no longer compiles. metadata(..) stays fallible (Result<Builder, serde_json::Error>) via a bon with closure. NewBalanceLimitBuilder::always_active() consumes the builder and is only callable while start is unset. NewTransaction.correlation_id defaults to id.to_string() via a bon default instead of the custom id setter. XBuilder type names still exist but are now generic over bon's typestate and have no Default; X::builder() is the only entry point.

Tests: the runtime fails_when_mandatory_fields_are_missing / fails_when_missing_required_fields unit tests (NewAccount, NewJournal, NewTransaction, NewTxTemplate, NewEntry) are removed because the case is now a compile error (agreed with Justin). New unit tests pin that invalid CEL, a mistyped param default, and a config with zero or two connection sources are still rejected. All other test edits are mechanical call-shape migrations.

Downstream call-site changes

From the lana-bank checkout (48b17701e5, worktrees and target/ excluded). Counts are X::builder() call sites / files.

Drop .expect(..) / .unwrap() / ? after build() (now infallible):

• NewAccount — 18 sites / 12 files (e.g. core/accounting/src/chart_of_accounts/{chart_node/entity.rs,ledger/closing_balances.rs}, core/{alt-credit,exchange,fx,holding,payments}/src/**/ledger/mod.rs, tests in core/accounting, core/credit/collection, core/fx, core/exchange)
• NewAccountSet — 18 / 12 (core/accounting/src/{balance_sheet,profit_and_loss,trial_balance,chart_of_accounts}/…, core/alt-credit/src/credit_facility_product/ledger/mod.rs, core/holding/src/ledger.rs, tests in core/{accounting,alt-credit,credit,deposit,exchange})
• NewJournal — 9 / 9 (core/accounting/src/journal/service.rs, core/payments/src/ledger/mod.rs, test helpers in 7 crates)
• NewTxTemplate — 26 / 26 (every template file under core/{accounting,alt-credit,exchange,fx,holding,payments}/src/**/templates* plus 5 test files)
• AccountSetUpdate — 1 / 1: core/accounting/src/chart_of_accounts/ledger/mod.rs:113, let mut update_values = AccountSetUpdate::default(); update_values.metadata(..).expect(..) → AccountSetUpdate::builder().metadata(..).expect(..).build()

Error type changes (fallible build() stays; only matters where the error is named or converted): NewTxTemplateEntry 66/26, NewTxTemplateTransaction 27/27, NewParamDefinition 123/17, NewVelocityControl 6/6, NewVelocityLimit 9/6, NewLimit 9/6, NewBalanceLimit 9/6, CalaLedgerConfig 23/18. Non-test lana code uses .expect(..) on all of these; the ? sites are tests returning anyhow::Result, which accept the new errors unchanged. No lana file names any cala XBuilderError, XBuilder or *UpdateValues type, and no From/#[from] impl wraps them, so no error-enum changes are needed.

let mut builder patterns (2 sites, both NewTxTemplateEntry with an optional metadata): core/exchange/src/ledger/templates/settle_internal_bitcoin_purchase.rs:221 and core/holding/src/template.rs:173 → a single chain with .maybe_metadata(opt).

Unchanged for lana: .always_active() (9 sites, all inline in chains), .metadata(..) on New* builders (4 sites, still Result), JobSvcConfig, default_expr.

Test plan

Gate output is in the task report: cargo fmt, cargo clippy --workspace --all-targets --all-features -D warnings, nix flake check (fmt, clippy, audit, deny), nix run .#nextest (nextest + doctests + docs, local Postgres), fuzz crate cargo check --all-targets, and make check-event-schemas (no schema diff). No .sqlx changes (no query changes).

🤖 Generated with Claude Code

---

Note

Medium Risk
Wide breaking API change to every entity/template builder and a major es-entity bump; behavior should match prior validation but downstream crates must update call sites and error types.

Overview
Replaces derive_builder with bon across the workspace and bumps es-entity 0.20.0, job 0.18.5, and obix 0.14.4 (lockfiles updated, including fuzz). Entity hydration for accounts, journals, templates, etc. now uses X::hydrator() from es-entity 0.20 instead of per-entity derive_builder on EsEntity types.

Builder API shifts: many New* / update types use bon::Builder with compile-time required fields (infallible .build()); validated constructors (CalaLedgerConfig, NewParamDefinition, tx-template CEL pieces, velocity limits) return typed errors (InvalidCalaLedgerConfig, InvalidParamDefinition, CelParseRejection) instead of stringly derive_builder validation. Update DTOs are renamed/consolidated (e.g. AccountUpdate replaces AccountUpdateValues). Optional fields gain maybe_* setters; metadata uses bon with closures.

Tests, examples, and perf crates are migrated mechanically (drop .unwrap() on infallible builds, chain builders instead of let mut).

^{Reviewed by Cursor Bugbot for commit c6a36f2. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

📊 Performance Report

Commit: c6a36f2
Updated: 2026-10-22 14:55:20 UTC

Cala Performance Benchmark Results (non-representative)

Criterion Benchmark Results (single-threaded)

Benchmark	Time per Run	Throughput	% vs Baseline
post_simple_transaction	2.617ms	381 tx/s	0 (baseline)
post_backdated_transaction_with_effective_balances_365_days	110.791ms	9 tx/s	-4132.0%
post_multi_layer_transaction	2.986ms	334 tx/s	-14.0%
post_simple_transaction_with_effective_balances	3.081ms	324 tx/s	-17.0%
post_simple_transaction_with_skipped_velocity	2.771ms	360 tx/s	-5.0%
post_simple_transaction_with_velocity	3.675ms	272 tx/s	-40.0%
post_simple_transaction_with_hit_velocity	1.788ms	558 tx/s	+31.0%
post_simple_transaction_with_one_account_set	3.161ms	316 tx/s	-20.0%
post_simple_transaction_with_five_account_sets	3.769ms	265 tx/s	-43.0%
post_simple_transaction_with_ec_account_set	3.678ms	271 tx/s	-40.0%

Load Testing Results (parallel-execution)

Scenario	tx/s
1 parallel	230.08
2 parallel	364.69
5 parallel	502.82
10 parallel	464.62
20 parallel	474.93
2 contention	260.88
5 contention	251.60
2 acct_sets	198.22
5 acct_sets	198.08

---

Note: Performance results may vary based on system resources and database state.

Last updated by commit c6a36f2