Indexer: treat TimeCurve buyFor as Buy; enrich with ETH/USDM pay metadata

PlasticDigits commented

2026-04-26 07:51:39 +00:00

(Migrated from gitlab.com)

Context

Issue #65 added TimeCurve.buyFor (companion TimeCurveBuyRouter) so CL8Y can be paid from the router while the buyer receives the same Buy event and CHARM/WarBow accounting as a normal buy. Optional TimeCurveBuyRouter.BuyViaKumbaya logs pay kind and amounts for observability.

Goals

Parity with normal buys — Index buyFor-driven purchases so they appear in the same pipelines as today’s Buy events (same TimeCurve Buy log shape: buyer = participant, charmWad, amount, fees, timer fields, etc.). Do not drop or mis-attribute buys because msg.sender on the sale contract was the router.
Enrichment for multi-asset entry — Persist additional fields (or a sidecar row / JSON column) when the tx is attributable to TimeCurveBuyRouter.buyViaKumbaya (decode BuyViaKumbaya + same-tx Buy correlation by tx hash / log index):
- Pay asset class: ETH vs USDM (stable) vs CL8Y-only (unchanged).
- Where useful and decodable: gross CL8Y (already on Buy as amount), optional implied input from BuyViaKumbaya / swap logs if stable for analytics later.
Schema / API — Extend internal models and any public API that exposes buys so downstream surfaces can optionally show “paid via router (ETH/USDM)” without breaking existing CL8Y-only consumers (defaults / nulls for legacy rows).

Acceptance criteria

buyFor txs produce the same canonical buy records as buy (correct buyer, amount, CHARM, indexer invariants).
When BuyViaKumbaya is present in the same tx as Buy, store ETH vs USDM (and any agreed numeric fields) for later UX/analytics.
Tests: decoder round-trip and/or integration test with fixture logs (docs/testing/invariants-and-business-logic.md test map).
Document new fields in indexer docs / changelog as appropriate.

References

contracts/src/TimeCurve.sol — buyFor, Buy event
contracts/src/TimeCurveBuyRouter.sol — BuyViaKumbaya
indexer/ decoders and persist layer for TimeCurve
Related frontend issue: #66 — single-tx UX

## Context [Issue #65](https://gitlab.com/PlasticDigits/yieldomega/-/issues/65) added `TimeCurve.buyFor` (companion `TimeCurveBuyRouter`) so CL8Y can be paid from the router while the **buyer** receives the same `Buy` event and CHARM/WarBow accounting as a normal `buy`. Optional `TimeCurveBuyRouter.BuyViaKumbaya` logs pay kind and amounts for observability. ## Goals 1. **Parity with normal buys** — Index `buyFor`-driven purchases so they appear in the same pipelines as today’s `Buy` events (same `TimeCurve` `Buy` log shape: `buyer` = participant, `charmWad`, `amount`, fees, timer fields, etc.). Do not drop or mis-attribute buys because `msg.sender` on the sale contract was the router. 2. **Enrichment for multi-asset entry** — Persist **additional** fields (or a sidecar row / JSON column) when the tx is attributable to `TimeCurveBuyRouter.buyViaKumbaya` (decode `BuyViaKumbaya` + same-tx `Buy` correlation by tx hash / log index): - Pay asset class: **ETH** vs **USDM** (stable) vs **CL8Y-only** (unchanged). - Where useful and decodable: **gross CL8Y** (already on `Buy` as `amount`), optional **implied input** from `BuyViaKumbaya` / swap logs if stable for analytics later. 3. **Schema / API** — Extend internal models and any public API that exposes buys so **downstream surfaces can optionally show** “paid via router (ETH/USDM)” without breaking existing CL8Y-only consumers (defaults / nulls for legacy rows). ## Acceptance criteria - [ ] `buyFor` txs produce the same **canonical buy** records as `buy` (correct buyer, `amount`, CHARM, indexer invariants). - [ ] When `BuyViaKumbaya` is present in the same tx as `Buy`, store **ETH vs USDM** (and any agreed numeric fields) for later UX/analytics. - [ ] Tests: decoder round-trip and/or integration test with fixture logs ([`docs/testing/invariants-and-business-logic.md`](docs/testing/invariants-and-business-logic.md) test map). - [ ] Document new fields in indexer docs / changelog as appropriate. ## References - `contracts/src/TimeCurve.sol` — `buyFor`, `Buy` event - `contracts/src/TimeCurveBuyRouter.sol` — `BuyViaKumbaya` - `indexer/` decoders and persist layer for TimeCurve - Related frontend issue: [#66](https://gitlab.com/PlasticDigits/yieldomega/-/issues/66) — single-tx UX

PlasticDigits commented

2026-04-26 07:51:46 +00:00

(Migrated from gitlab.com)

mentioned in issue #66

PlasticDigits commented

2026-04-26 07:51:48 +00:00

(Migrated from gitlab.com)

Companion: frontend single-tx buyViaKumbaya wiring — https://gitlab.com/PlasticDigits/yieldomega/-/issues/66

Companion: frontend single-tx `buyViaKumbaya` wiring — https://gitlab.com/PlasticDigits/yieldomega/-/issues/66

PlasticDigits commented

2026-04-26 07:51:50 +00:00

(Migrated from gitlab.com)

Cross-link: Replace the placeholder “frontend issue” in the description with issue #66.

**Cross-link:** Replace the placeholder “frontend issue” in the description with issue [#66](https://gitlab.com/PlasticDigits/yieldomega/-/issues/66).

PlasticDigits commented

2026-04-26 08:15:54 +00:00

(Migrated from gitlab.com)

changed the description

PlasticDigits commented

2026-04-26 08:17:01 +00:00

(Migrated from gitlab.com)

mentioned in commit 5eaf345725

mentioned in commit 5eaf34572578b70d7ab35624f8070ca3eb27a400

PlasticDigits commented

2026-04-26 08:17:01 +00:00

(Migrated from gitlab.com)

mentioned in commit 2a89c4941e

mentioned in commit 2a89c4941eda9f7f5c589e2d8786d0005bbbd050

PlasticDigits commented

2026-04-26 08:17:34 +00:00

(Migrated from gitlab.com)

@brouie Please verify the indexer fix for #67 (merged to main in 5eaf345).

What changed

BuyViaKumbaya is decoded and stored in idx_timecurve_buy_router_kumbaya (idempotent on tx_hash, log_index).
ADDRESS_REGISTRY JSON may include optional TimeCurveBuyRouter so eth_getLogs ingests router events (empty string = omit from filter).
GET /v1/timecurve/buys (x-schema-version 1.8.0) left-joins matching rows and adds optional fields when the same tx has a correlated sidecar: entry_pay_asset (eth | stable) and router_attested_gross_cl8y (from the router event). Legacy / CL8Y-only buys omit these keys.
Reorg rollback deletes the new table. Integration + decoder round-trip tests cover the new path.
Docs: docs/integrations/kumbaya.md, docs/indexer/design.md, docs/testing/invariants-and-business-logic.md, skills/README.md, .cursor/skills/yieldomega-guardrails/SKILL.md. Issue description updated to link #66 for the frontend companion.

Checklist for you

Run migrations on a staging DB and confirm idx_timecurve_buy_router_kumbaya exists.
Set TimeCurveBuyRouter in ADDRESS_REGISTRY_PATH to the deployed router; restart ingestion; confirm router logs appear in the table after a buyViaKumbaya tx.
Confirm GET /v1/timecurve/buys returns entry_pay_asset / router_attested_gross_cl8y for those txs and omits them for ordinary buy txs.
Confirm buyer on buy rows matches the participant (unchanged invariant — sourced from TimeCurve Buy).
CI / cd indexer && cargo test green with YIELDOMEGA_PG_TEST_URL where you use Postgres integration.

Leaving the issue open for your sign-off.

@brouie Please verify the indexer fix for **#67** (merged to `main` in 5eaf345). ## What changed - **`BuyViaKumbaya`** is decoded and stored in **`idx_timecurve_buy_router_kumbaya`** (idempotent on `tx_hash, log_index`). - **`ADDRESS_REGISTRY`** JSON may include optional **`TimeCurveBuyRouter`** so **`eth_getLogs`** ingests router events (empty string = omit from filter). - **`GET /v1/timecurve/buys`** (`x-schema-version` **1.8.0**) left-joins matching rows and adds optional fields when the same tx has a correlated sidecar: **`entry_pay_asset`** (`eth` \| `stable`) and **`router_attested_gross_cl8y`** (from the router event). Legacy / CL8Y-only buys omit these keys. - **Reorg rollback** deletes the new table. **Integration + decoder round-trip** tests cover the new path. - Docs: **`docs/integrations/kumbaya.md`**, **`docs/indexer/design.md`**, **`docs/testing/invariants-and-business-logic.md`**, **`skills/README.md`**, **`.cursor/skills/yieldomega-guardrails/SKILL.md`**. Issue description updated to link **#66** for the frontend companion. ## Checklist for you - [ ] Run migrations on a staging DB and confirm **`idx_timecurve_buy_router_kumbaya`** exists. - [ ] Set **`TimeCurveBuyRouter`** in **`ADDRESS_REGISTRY_PATH`** to the deployed router; restart ingestion; confirm router logs appear in the table after a **`buyViaKumbaya`** tx. - [ ] Confirm **`GET /v1/timecurve/buys`** returns **`entry_pay_asset`** / **`router_attested_gross_cl8y`** for those txs and omits them for ordinary **`buy`** txs. - [ ] Confirm **`buyer`** on buy rows matches the participant (unchanged invariant — sourced from **`TimeCurve` `Buy`**). - [ ] CI / `cd indexer && cargo test` green with **`YIELDOMEGA_PG_TEST_URL`** where you use Postgres integration. Leaving the issue **open** for your sign-off.

PlasticDigits commented

2026-04-26 08:23:08 +00:00

(Migrated from gitlab.com)

mentioned in commit 3d45e3b90d

mentioned in commit 3d45e3b90dcf6f9e75b4ec3f17f633c0e2c9b3f0

Brouie commented

2026-04-27 02:46:33 +00:00

(Migrated from gitlab.com)

mentioned in issue #72

Brouie commented

2026-04-27 02:55:13 +00:00

(Migrated from gitlab.com)

@PlasticDigits source + Postgres-roundtrip verification on 5eaf345.

Source review

Decoder (src/decoder.rs):

L141: BuyViaKumbaya event ABI
L278: DecodedEvent::TimeCurveBuyRouterBuyViaKumbaya enum variant
L705-708: signature_hash dispatch and decode_log
L922: roundtrip test (round-trip AC item)

Persist (src/persist.rs L125-133):

INSERT INTO idx_timecurve_buy_router_kumbaya
ON CONFLICT idempotent on (tx_hash, log_index) PK

Reorg rollback (src/reorg.rs L119):

Table listed in delete chain

Config (src/config.rs L23-25):

timecurve_buy_router optional, empty string omits from filter

API (src/api.rs):

L22 SCHEMA_VERSION = '1.8.0'
L50 kumbaya_entry_pay_asset() helper maps pay_kind smallint to 'eth' / 'stable'
L111 header set to x-schema-version
L207-210 entry_pay_asset + router_attested_gross_cl8y as Option (omitted from JSON when null)
L254 SELECT pulls gross_cl8y::text AS router_attested_gross_cl8y from join
L314-318 row mapping into response DTO

Migration

Up SQL (migrations/20260426120000_timecurve_buy_router_kumbaya.up.sql):

CREATE TABLE IF NOT EXISTS idx_timecurve_buy_router_kumbaya
block_number, block_hash, tx_hash, log_index, contract_address, buyer (TEXT), charm_wad NUMERIC(78,0), gross_cl8y NUMERIC(78,0), pay_kind SMALLINT
PK (tx_hash, log_index), INDEX (tx_hash, buyer, charm_wad)

Down SQL: clean DROP TABLE IF EXISTS

Applied to real yieldomega_indexer DB on yieldomega-pg, table structure inspected via psql \d:

All columns + types match SQL spec
idx_timecurve_buy_router_kumbaya_pkey + idx_tcb_kumbaya_lookup both present

Tests

cd indexer && cargo test:

Unit: 15/15 passing (lib + binary)
Integration: postgres_stage2_persist_all_events_and_rollback_after 1/1 passing against real Postgres at YIELDOMEGA_PG_TEST_URL=postgres://yieldomega:password@127.0.0.1:5433/yieldomega_indexer
The integration test exercises BuyViaKumbaya insert at tests/integration_stage2.rs L362, count assertion at L630/726/861, correlation walk at L319, and the reorg-rollback delete path

Deferred until #72 lands (not closeable today)

Live router log ingestion — TimeCurveBuyRouter not deployable on Anvil because TimeCurve.sol exceeds EIP-170 24KB cap (28,454 bytes), filed as #72 LAUNCH-BLOCKER this morning
End-to-end buyViaKumbaya tx producing eth/stable rows + correlated buy
API response on real data via GET /v1/timecurve/buys with x-schema-version: 1.8.0 header

Source layer is solid. Will close once #72 unblocks the deploy and I can ingest a real router event end-to-end.

Issue stays open.

@PlasticDigits source + Postgres-roundtrip verification on 5eaf345. ## Source review Decoder (src/decoder.rs): - L141: BuyViaKumbaya event ABI - L278: DecodedEvent::TimeCurveBuyRouterBuyViaKumbaya enum variant - L705-708: signature_hash dispatch and decode_log - L922: roundtrip test (round-trip AC item) Persist (src/persist.rs L125-133): - INSERT INTO idx_timecurve_buy_router_kumbaya - ON CONFLICT idempotent on (tx_hash, log_index) PK Reorg rollback (src/reorg.rs L119): - Table listed in delete chain Config (src/config.rs L23-25): - timecurve_buy_router optional, empty string omits from filter API (src/api.rs): - L22 SCHEMA_VERSION = '1.8.0' - L50 kumbaya_entry_pay_asset() helper maps pay_kind smallint to 'eth' / 'stable' - L111 header set to x-schema-version - L207-210 entry_pay_asset + router_attested_gross_cl8y as Option<String> (omitted from JSON when null) - L254 SELECT pulls gross_cl8y::text AS router_attested_gross_cl8y from join - L314-318 row mapping into response DTO ## Migration Up SQL (migrations/20260426120000_timecurve_buy_router_kumbaya.up.sql): - CREATE TABLE IF NOT EXISTS idx_timecurve_buy_router_kumbaya - block_number, block_hash, tx_hash, log_index, contract_address, buyer (TEXT), charm_wad NUMERIC(78,0), gross_cl8y NUMERIC(78,0), pay_kind SMALLINT - PK (tx_hash, log_index), INDEX (tx_hash, buyer, charm_wad) Down SQL: clean DROP TABLE IF EXISTS Applied to real yieldomega_indexer DB on yieldomega-pg, table structure inspected via psql \d: - All columns + types match SQL spec - idx_timecurve_buy_router_kumbaya_pkey + idx_tcb_kumbaya_lookup both present ## Tests cd indexer && cargo test: - Unit: 15/15 passing (lib + binary) - Integration: postgres_stage2_persist_all_events_and_rollback_after 1/1 passing against real Postgres at YIELDOMEGA_PG_TEST_URL=postgres://yieldomega:password@127.0.0.1:5433/yieldomega_indexer - The integration test exercises BuyViaKumbaya insert at tests/integration_stage2.rs L362, count assertion at L630/726/861, correlation walk at L319, and the reorg-rollback delete path ## Deferred until #72 lands (not closeable today) - Live router log ingestion — TimeCurveBuyRouter not deployable on Anvil because TimeCurve.sol exceeds EIP-170 24KB cap (28,454 bytes), filed as #72 LAUNCH-BLOCKER this morning - End-to-end buyViaKumbaya tx producing eth/stable rows + correlated buy - API response on real data via GET /v1/timecurve/buys with x-schema-version: 1.8.0 header Source layer is solid. Will close once #72 unblocks the deploy and I can ingest a real router event end-to-end. Issue stays open.

Brouie commented

2026-04-27 04:03:41 +00:00

(Migrated from gitlab.com)

mentioned in issue #75

Brouie commented

2026-04-27 04:04:03 +00:00

(Migrated from gitlab.com)

@PlasticDigits per today's process update (file follow-ups, close originals): live ingestion items moved to #75 tracker. Source + real-Postgres roundtrip verification (see prior note on this issue) stands.

Closing #67. Live walkthrough proceeds via #75 once a buyViaKumbaya tx is captured (likely via #74 browser walkthrough).

@PlasticDigits per today's process update (file follow-ups, close originals): live ingestion items moved to #75 tracker. Source + real-Postgres roundtrip verification (see prior note on this issue) stands. Closing #67. Live walkthrough proceeds via #75 once a buyViaKumbaya tx is captured (likely via #74 browser walkthrough).

Brouie (Migrated from gitlab.com) closed this issue

2026-04-27 04:04:04 +00:00

PlasticDigits commented

2026-04-27 07:24:01 +00:00

(Migrated from gitlab.com)

mentioned in issue #83

PlasticDigits commented

2026-04-28 03:21:40 +00:00

(Migrated from gitlab.com)

mentioned in issue #84

PlasticDigits commented

2026-05-05 08:35:15 +00:00

(Migrated from gitlab.com)

mentioned in issue #156

Rows
Columns

Indexer: treat TimeCurve buyFor as Buy; enrich with ETH/USDM pay metadata #67

Context

Goals

Acceptance criteria

References

What changed

Checklist for you

Source review

Migration

Tests

Deferred until #72 lands (not closeable today)