GraphAI Lite

GraphAI Lite is a light-weight version of GraphAI, the declarative data-flow programming framework that makes it easy to develop complex applications utilizing multiple asynchronous calls and their concurrent executions.

GraphAI Lite consists of a set of small TypeScript libraries, allowing developers to leverage data-flow programming and eliminate the need to manage complex states involving asynchronous calls and concurrent executions.

Usage

While async/await has simplified calling asynchronous functions, everything is executed sequentially, even when the tasks are independent.

const ExecuteAtoF = async () => {
  const a = await FuncA();
  const b = await FuncB();
  const c = await FuncC();
  const d = await FuncD(a, b);
  const e = await FuncE(b, c);
  return FuncF(d, e);
};

To execute independent asynchronous tasks concurrently, you need to use Promise.all. However, fully optimizing this can lead to hard-to-maintain code.

For example, the code below is slightly optimized but not fully optimized (FuncD needs to wait until FuncC is done even though there is no dependency, and FuncE needs to wait for FuncA).

const ExecuteAtoF = async () => {
  const [a, b, c] = await Promise.all([FuncA(), FuncB(), FuncC()]);
  const [d, e] = await Promise.all([FuncD(a, b), FuncE(b, c)]);
  return FuncF(d, e);
};

The computed function of GraphAI Lite is a thin wrapper around Promise.all, enabling data-flow style programming. You just need to specify the dependencies of various tasks (nodes) and let the system figure out the appropriate execution order.

import { computed } from '@receptron/graphai_lite';

const ExecuteAtoF = async () => {
  const nodeA = FuncA();
  const nodeB = FuncB();
  const nodeC = FuncC();
  const nodeD = computed([nodeA, nodeB], FuncD);
  const nodeE = computed([nodeB, nodeC], FuncE);
  const nodeF = computed([nodeD, nodeE], FuncF);
  return nodeF;
};

Below is the data-flow diagram represented by the code above (the graph is visible only on Github).

flowchart LR
 nodeA --> nodeD
 nodeB --> nodeD
 nodeB --> nodeE
 nodeC --> nodeE
 nodeD --> nodeF
 nodeE --> nodeF

Conductor

The Conductor class of GraphAI Lite allows developers to log the executions of those asynchronous tasks.

import { Conductor } from '@receptron/graphai_lite';

const ExecuteAtoF = async (conductor: Conductor) => {
  const nodeA = conductor.computed([], FuncA, { name: "nodeA" });
  const nodeB = conductor.computed([], FuncB, { name: "nodeB" });
  const nodeC = conductor.computed([], FuncC, { name: "nodeC" });
  const nodeD = conductor.computed([nodeA, nodeB], FuncD, { name: "nodeD" });
  const nodeE = conductor.computed([nodeB, nodeC], FuncE, { name: "nodeE" });
  const nodeF = conductor.computed([nodeD, nodeE], FuncF, { name: "nodeF" });
  conductor.result = {
    f: await nodeF
  };
};

const main = async () => {
  const conductor = new Conductor({ verbose:true, recordInputs: true, recordOutput: true });
  await ExecuteAtoF(conductor);
  console.log(conductor.logs);
  console.log(conductor.result);
}