Differences (The Graph vs. Envio)

warning

If you're looking to support both subgraphs (The Graph) and indexers (Envio) into your apps, see the "Different Results (*)" 📌 section below to understand how results from these two APIs may differ for the same entities.

General

In contrast to The Graph, Envio offers a slightly different Developer Experience (DX) in terms of indexer implementation. While the query specifics don't change a lot for consumers, here are a few items to consider before developing on top of the Sablier indexers/subgraphs:

Querying Language

Both solutions create a GraphQL API for consumers to read data from. The Graph uses a customized subset of GraphQL operations, which makes is impossible to use the same queries between both indexers and subgraphs, whereas Envio deploys a GraphQL API over a Postgres DB. Here are some examples:

Example	The Graph	Envio
Pagination	first, skip	limit, offset
Filter by id	stream(id)	Stream(where: {id: {_eq: 1}})
Single vs. Plural	stream(id){}, streams{}	Stream(where...){} for both
Nested items	campaigns{ id, asset: {id, symbol}}	Campaign{ asset_id, asset: {id, symbol}}
Filter similar items	assets(id: $id)	Asset(where: {id: {_iregex: $id}})

Notes:

• For the 3rd example, querying for a single item vs. a collection will have separate keywords for The Graph. With Envio, you'll have to identify within the app itself if the query is supposed to yield one or multiple items. As you may tell, with Envio you'll query for Stream (capitalized), while with The Graph you'll query for stream or streams.
• For the 4th example, Envio Indexers will yield the contents of the name of an object with ({}), while the <name>_id label will ask for the identifier of the object (e.g., asset vs asset_id). Keep this in mind for "where" filtering.

Handler Language

• The Graph: uses Assembly Script
• Envio: Sablier uses TypeScript by default, but you can also use JavaScript or Rescript

With Envio, one important mental model is in the concept of loaders vs. handlers. To optimize querying speeds, Envio will ask you to write 2 methods: (i) a loader, where you can express which existing entities you're expecting to use and mutate, and (ii) a handler where you manage these entities and/or create new ones.

Example: In the case of a Withdraw event, we weill pre-load the Stream entity and maybe the Watcher, while in the second part we'll handle the creation of a new Action of type Withdraw, we'll update the Stream's withdrawnAmount and increase the Watcher's index.

Specifics

Initializer Contracts

We've architected this indexer around a set of pre-configured contracts.

Similar to The Graph, we start by pre-configuring a set of contracts. While Envio's indexer doesn't have the same requirement of pre-configuring contracts to listen to, we'll keep this feature to ensure we can query against those entities, even if they'll be empty at start.

We'll ensure contracts have been initialized (see the watcher.ts helper) by making a call against the initializer at the start of each method. It should only come into play within the create handlers.

Versioning and Lockup Flavors

While for The Graph's subgraphs we track flavor-first (see subgraph.template.yaml for the configuration of SablierV2LockupLinear and SablierV2LockupDynamic), for Envio's indexers we'll have a version-first approach.

Therefore, LockupLinear and LockupDynamic will be bundled under the same Lockup<Version> contract tracker (see ./config.template.mustache). Different versions of the protocol will be tracked separately, which is why we have Lockup_V20 (v2.0) and Lockup_V21 (v2.1) in our configuration. Later on, inside the handler logic, we'll separate contracts by flavor.

Envio: Contract configuration tree (version-first)

└── contracts/
    ├── LockupV20/
    │   ├── event: CreateLockupDynamicStream
    │   └── event: CreateLockupLinearStream
    └── LockupV21/
        ├── event: CreateLockupDynamicStream
        └── event: CreateLockupLinearStream

The Graph: Contract configuration tree (flavor-first)

└── contracts/
    ├── LockupDynamic/
    │   ├── event: CreateLockupDynamicStreamV20
    │   └── event: CreateLockupDynamicStreamV21
    └── LockupLinear/
        ├── event: CreateLockupLinearStreamV20
        └── event: CreateLockupLinearStreamV21

Contract Calls

Outside of data provided directly by the event arguments, one can access additional information through contract calls. While The Graph provides a dedicated API (binding an ABI to an address, calling contract.try_method()), Envio is less opinionated. A good example of contract calls can be found in our helpers/asset.ts where we use viem and a custom caching mechanism to search for asset details.

Different Results (*)

Due to the cross-chain indexing aspect, entities in Envio will need to have a chainId suffix attached to them to ensure they're unique across the board. At the same time, there are some minor features missing, which will cause some differences listed below.

• For Envio indexers, some entities will have different identifiers (from what The Graph's subgraph have):
  1. protocol-envio: the Action, Asset, Batch, Batcher, Contract have a -chainId appended to the ID
  2. merkle-envio: the Action, Asset and Factory have a -chainId appended to the ID

tip

To avoid writing separate systems when assigning variables to queries, you can use slightly different filters. For example, given the different ids of an Asset (address in The Graph and address-chainId in Envio) you can query for certain assets with:

1. asset(id: $assetId) for The Graph
2. Asset(where: {id: {_iregex: $assetId}})

With assetId in both cases being assigned to the Asset's address.