# UMA SDK Tables

These tables give you an API for dealing with data of a certain type which can be keyed into tables. Tables should
give you useful methods for manipulating and querying data in each table. Tables rely on [stores](../stores/README.md).

## Tables

- [base](./base/README.md) - Some basic table functionality which most tables can inherit from
- [blocks](./blocks/README.md) - For storing blocks
- [emps](./emps/README.md) - For storing emp contract state
- [erc20s](./erc20s/README.md) - For storing erc20 contract state
- [historical-prices](./historical-prices/README.md) - For storing historical prices.

## Relationship to Stores

Tables consume a store, which expose a CRUD like way to manipulate data in a persistent store or cache. Tables will consume
and wrap store functionality and potentially add data specific ways to manipulate that particular table. Each table should
represent a type of data, but they may expose multiple implementations based on the type of store used.

## Queries

Tables may require data store specific ways of querying data, ie ability to access underlying database client in many cases.

## Adding a Table

Create a new folder named after the type of data you are storing in the tables directory with the following files:

- README.md: Add some notes on what the table is for and any stores it can use
- index.ts: Expose your tables based on the name type of store it uses, for example `export {default as JsMap} from './js-map'`
- \${store-type}.ts: Create a table file named for the store its compatible with, for example `js-map.ts`
- \${store-type}.test.ts: Any tests you want to run

See the [blocks table](./blocks/README.md) as an example.

## Usage Considerations

These tables are designed to work in such a way that data is queried and updated in the application layer
rather than in the database layer. This is a subtle distinction, but in some usecases this may cause race conditions
during times where many updates may be happening concurrently ( queried state within table functions may be inconsitent with database state).
If this applies to your use case, you may need to write a custom table using the database driver directly or
order your mutations such that they execute serially.