ArborVote Indexer

An Envio HyperIndex indexer for the ArborVote contract. It folds the contract's event stream into queryable domain entities - Debate, Argument (the tree, with market reserves and tally impact), Participant (token balances), Position (share holdings), and the append-only Stake/Redemption histories - so clients can read a whole debate in one GraphQL query instead of RPC-traversing the tree leaf by leaf.

Every contract event carries the resulting state (reserves move additively, payouts arrive pre-rounded), so the handlers mirror the debate without redoing any market math. The event set is documented in contracts/src/interfaces/IArborVote.sol.

Develop

just install   # npm install
just codegen   # regenerate types from config.local.yaml + schema.graphql
just test      # type-check + handler tests (in-memory, no database needed)
just dev       # run against the local anvil chain (docker: postgres + hasura)

Two configs: config.yaml (the default) targets Base Sepolia for the Envio hosted service; config.local.yaml targets the local anvil chain. The just dev recipes point envio at the local config (--config config.local.yaml / ENVIO_CONFIG), so plain envio commands still default to the hosted config.

just dev expects the frontend dev stack (just dev-anvil in frontend/) to be running: it indexes chain 31337 at http://127.0.0.1:8545 from block 0. Every dev-anvil run writes its deployment's address into this repo's .env (ENVIO_ARBORVOTE_ADDRESS), which config.local.yaml interpolates - so the index follows the newest deployment even when a reused anvil moves the contract to a fresh nonce. Because the chain is ephemeral, so is the index: just dev wipes and re-indexes from block 0 on every start (local chains are small; this takes seconds). Hasura's GraphQL console comes up on http://localhost:8090 (moved off 8080, which the dev kubo gateway occupies - the recipe pins both the container port and envio's metadata endpoint there; local password testing).

The handler tests simulate event streams against an in-memory indexer - the lifecycle test replays the same numbers as the contract unit tests (seed at 80%, rate down, redeem at a profit), asserting that the folded entities match the chain exactly.

Hosted service (Base Sepolia)

The default config.yaml indexes the Base Sepolia deployment (chain 84532) via HyperSync - no RPC endpoint needed. On envio.dev's hosted service, with this repo connected:

1. Leave the deployment's config file at the default config.yaml (branch main, root directory .); the hosted service picks it up automatically.
2. The ArborVote address and deployment block are already in the config. To point at a redeploy without editing it, set ENVIO_ARBORVOTE_ADDRESS in the environment variables tab instead.
3. Optionally set ENVIO_PIN_IPFS_API once a pinning node exists (see below).

Every push to main redeploys the indexer. The deployment's GraphQL endpoint (shown in the hosted app) becomes the frontend's VITE_INDEXER_URL.

Production pinning backstop

Argument texts are IPFS raw-leaves blocks whose sha-256 digests are public on-chain. When ENVIO_PIN_IPFS_API is set (e.g. http://127.0.0.1:5001), the indexer re-pins every content digest it sees - debate theses, added and altered arguments - on that kubo-compatible node, so content availability never depends on the authoring client alone (see the frontend README, "Production pinning strategy"). Pinning is idempotent and best-effort: a failure never stalls indexing, and the next resync retries.

Prerequisites

• Node.js v22+
• Docker (only for just dev, not for tests)