Install

npm i @rebirthdb/rebirthdb-ts

or

yarn add @rebirthdb/rebirthdb-ts

Import

// if you support import
import { r } from '@rebirthdb/rebirthdb-ts';
// if you dont
const { r } = require('@rebirthdb/rebirthdb-ts');

Initialize

// in an async context
// if you want to initialize a connection pool
await r.connectPool(options);
// if you want to initialize a single connection
const conn = await r.connect(options);

STATUS:

• Fully working typescript driver!
• Rebuilt from scrach using the latest ES/TS features for readability and maintainability
• Drop-in replacement for rethinkdbdash with only some minor changes

CHANGES FROM RETHINKDBDASH

• Importing property instead of entire library: const {r} = require('rebirthdbts') or import {r} from 'rebirthdbts' instead of const r = require('rethinkdbdash')(options)
• No top level initialization, initializing a pool is done by await r.connectPool()
• No { cursor: true } option, for getting a cursor use .getCursor(runOptions) instead of .run(runOptions)
  • .run() will coerce streams to array by default feeds will return a cursor like rethinkdbdash
• Uses native promises instead of bluebird
• A cursor is already a readable stream, no need for toStream()
• A readable stream is already an async iterator in node 10 no need for .asyncIterator()
• In connction pool, reusing open connections that already run queries instead of making queries wait for a connection when max connections exceeded.
• Integrated fully encompasing type definitions

NEW FEATURES

• serialize / deserialize. You can store the query as a string like this const serializedQuery = r.table(...).filter(...).map(...).serialize() and get it like this r.deserialize(serializedQuery).run() or even r.deserialize<RStream>(serializedQuery).reduce(...).run() the serialized query is a normal string so you can store it in the DB. No need for ugly workarounds like .toString and eval anymore. Also the serialized query is the actual JSON that gets sent to the server so it should be cross-language compatible if any other driver cares to implement it.

DROPPING SUPPORT:

• Support node < 8
• Support callbacks.
• Support using .then() directly on a query (optionalRun), it can confuse users that queries are promises leading to false assumptions:
  • Queries are not promises since they are not eagerly evaluated and therefore they can:
    • .run() as many times as you want (promises run only once and return the same value without running other times)
    • be stored for future evaluation (promises run as you create them)
• Support browsers (Unless it's the only demand of making this driver used instead of rethinkdbdash)
• support for r.row you can use row => row instead. (may add support in the future)
• Support write streams (Does anyone uses it? will add it if its a popular demand)
• Multiple connection pools (if someone has a good usecase I'll support it)

TASKS REMAINING BEFORE RELEASE:

Priority - medium

• Publish in NPM
• Go through all type definitions and fix according to config (maybe use https://github.com/rethinkdb/rethinkdb/blob/3edaeceb71c2caf1203025a752f61786364528ed/drivers/java/term_info.json)

Priority - low

• New API tests (write hooks and bit ops)
• Supporting implicit var (r.row)
  • Use a lambda expression instead (row => row)

Priority - none

• Don't throw on r.expr(NaN) (only on r.expr(NaN).run()). Why? (test: r.expr should not NaN if not run)
• Client backtraces - because of the above NaN values throw in the right callstack (line + col) so backtraces are not nessesary
• Function suggestions fails ("noReplyWait should throw")
  • Typescript can help users better understand the right function names
• Suggesting optional arguments available options fails ("run should throw on an unrecognized argument")
  • still showing wrong argument exception + backtrace
  • Typescript can help users better understand the right params
• Not supporting certain top-level functions ("r.wait should throw")
  • Every not top level function can be translated to top level function by adding the query-term as the first arg: r.table('test').reconfigure({...}) -> r.reconfigure(r.table('test'), { ... })
  • This support will help make use of the future |> functional operator:
    • this.table('test') |> r.reconfigure(#, {})
• Throw special error if a top-level function is not defined on a term ("js is not defined after a term")
  • Throwing the standard TypeError: xxx is not a function
• Supporting .asyncIterator()
  • Cursor is a stream reader which is an async iterator by default in node 10
• Special r.time arity check (the current check is enough)
• Error message mismatch
• r.and() r.or() with no arguments