Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

Chain some async tasks in fp-ts retaining every task's result

In fp-ts, I'm trying to chain some potentially failing async tasks together with TaskEither but I need to use the results from intermediate tasks later on down the chain.

In this example:

const getFoo = (a: string): Promise<Foo> => {};
const getBar = (foo: Foo): Promise<Bar> => {};
const mkFooBar = (foo: Foo, bar: Bar): Promise<FooBar> => {};

const async main1: Promise<FooBar> => {
  const a = "a";
  const foo = await getFoo(a);
  const bar = await getBar(foo);
  const fooBar = await mkFooBar(foo, bar);

  return Promise.resolve(fooBar);
};

const main2: Promise<FooBar> => {
  const a = "a";

  return pipe(
    TE.tryCatch(() => getFoo(a), e => e),
    TE.chain(foo => TE.tryCatch(() => getBar(foo), e => e)),
    TE.chain(bar => TE.tryCatch(() => mkFooBar(??, bar), e => e))
  );
};

the main1 function is an async/await-style solution to this problem. What I'm trying to do is emulate something like this in a fp-ts chain-style. main2 is my attempt at this.

Because the async/await version introduces all the intermediate results into the local scope (i.e. foo and bar), it's easy to call mkFooBar which depends on both those results.

But in the fp-ts version, the intermediate results are trapped in the scope of each task.

The only way I can think to make this version work would be to make either the async functions themselves (i.e. getFoo and getBar) also return their arguments, or perhaps the TaskEither wrappers return the arguments so that they can then be passed on to the next function in the chain.

Would this be the correct way to do this? Or is there a simpler version which more closely resembles the async/await version?

like image 584
ironchicken Avatar asked Oct 25 '19 14:10

ironchicken


People also ask

What are asynchronous tasks and error handling in FP-TS?

This post will introduce two concepts in fp-ts: asynchronous tasks and error handling. Namely we will look at the Task, Either, and TaskEither types. Every asynchronous operation in modern Typescript is done using a Promise object. A task is a function that returns a promise which is expected to never be rejected.

What is an asynchronous task?

An asynchronous call. Something that can go wrong. In functional programming we use certain data types to deal with these types of complexity. In our case specifically: Reader - for dependency injection. Task - for asynchronous things. Either - for things that can go wrong.

What is a task in Python?

A task is a function that returns a promise which is expected to never be rejected. The type definition for task can be found below. Another way to define task is using a function type definition. Tasks are expected to always succeed but can fail when an error occurs outside our expectations.

What is a task in typescript?

Every asynchronous operation in modern Typescript is done using a Promise object. A task is a function that returns a promise which is expected to never be rejected. The type definition for task can be found below. Another way to define task is using a function type definition.


Video Answer


1 Answers

Depending on how many times you'll need to access the intermediate results in the following computation, I would suggest either using Do (an approximation of Haskell's do notation), or carry the intermediate results over via manual mapping.

Given:

import { pipe } from "fp-ts/function";
import * as TE from "fp-ts/TaskEither";

declare function getFoo(a: string): TE.TaskEither<unknown, Foo>;
declare function getBar(foo: Foo): TE.TaskEither<unknown, Bar>;
declare function mkFooBar(foo: Foo, bar: Bar): TE.TaskEither<unknown, FooBar>;

Example with Do:

function main2(): TE.TaskEither<unknown, FooBar> {
  return pipe(
    TE.Do,
    TE.bind("foo", () => getFoo("a")),
    TE.bind("bar", ({ foo }) => getBar(foo)),
    TE.chain(({ foo, bar }) => mkFooBar(foo, bar))
  );
}

Example mapping manually:

function main3(): TE.TaskEither<unknown, FooBar> {
  return pipe(
    getFoo("a"),
    TE.chain(foo =>
      pipe(
        getBar(foo),
        TE.map(bar => ({ foo, bar }))
      )
    ),
    TE.chain(({ foo, bar }) => mkFooBar(foo, bar))
  );
}
like image 69
Giovanni Gonzaga Avatar answered Nov 08 '22 23:11

Giovanni Gonzaga