A thenable to streamline a possibly sync / possibly async workflow.
yarn add value-or-promise
or npm install value-or-promise
Instead of writing:
function myFunction() {
const valueOrPromise = getValueOrPromise();
if (isPromise(valueOrPromise)) {
return valueOrPromise.then(v => onValue(v));
}
return onValue(valueOrPromise);
}
...write:
function myFunction() {
return new ValueOrPromise(getValueOrPromise)
.then(v => onValue(v))
.resolve();
}
When working with functions that may or may not return promises, we usually have to duplicate handlers in both the synchronous and asynchronous code paths. In the most basic scenario included above, using value-or-promise
already provides some code savings, i.e. we only have to reference doSomethingWithValue
once.
Things start to get even more beneficial when we add more sync-or-async functions to the chain.
Instead of writing:
function myFunction() {
const valueOrPromise = getValueOrPromise();
if (isPromise(valueOrPromise)) {
return valueOrPromise
.then(v => first(v))
.then(v => second(v));
}
const nextValueOrPromise = first(ValueOrPromise)
if (isPromise(nextValueOrPromise)) {
return nextValueOrPromise.then(v => second(v));
}
return second(nextValueOrPromise);
}
...write:
function myFunction() {
return new ValueOrPromise(getValueOrPromise)
.then(v => first(v))
.then(v => second(v))
.resolve();
}
Even with shorter chains, value-or-promise
comes in handy when managing errors.
Instead of writing:
function myFunction() {
try {
const valueOrPromise = getValueOrPromise();
if (isPromise(valueOrPromise)) {
return valueOrPromise
.then(v => onValue(v))
.catch(error => console.log(error));
}
const nextValueOrPromise = onValue(valueOrPromise);
if (isPromise(nextValueOrPromise)) {
return nextValueOrPromise
.catch(error => console.log(error));
}
return nextValueOrPromise;
} catch (error) {
console.log(error);
}
}
...write:
function myFunction() {
return new ValueOrPromise(getValueOrPromise)
.then(v => onValue(v))
.catch(error => console.log(error))
.resolve();
}
A simpler way of streamlining the above is to always return a promise.
Instead of writing:
function myFunction() {
const valueOrPromise = getValueOrPromise();
if (isPromise(valueOrPromise)) {
return valueOrPromise.then(v => onValue(v));
}
return onValue(valueOrPromise);
}
...or writing:
function myFunction() {
return new ValueOrPromise(getValueOrPromise)
.then(v => onValue(v))
.resolve();
}
...we could write:
function myFunction() {
return Promise.resolve(getValueOrPromise)
.then(v => onValue(v));
}
...but then we would always have to return a promise! If we are trying to avoid the event loop when possible, this will not suffice.
We can use ValueOrPromise.all(...)
analogous to Promise.all(...)
to create a new ValueOrPromise
object that will either resolve to an array of values, if none of the passed ValueOrPromise
objects contain underlying promises, or to a new promise, if one or more of the ValueOrPromise
objects contain an underlying promise, where the new promise will resolve when all of the potential promises have resolved.
For example:
function myFunction() {
const first = new ValueOrPromise(getFirst);
const second = new ValueOrPromise(getSecond);
return ValueOrPromise.all([first, second]).then(
all => onAll(all)
).resolve();
}
myFunction
with return a value if and only if getFirst
and getSecond
both return values. If either returns a promise, myFunction
will return a promise. If both getFirst
and getSecond
return promises, the new promise returned by myFunction
will resolve only after both promises resolve, just like with Promise.all
.
The value-to-promise
concept is by Ivan Goncharov.
Implementation errors are my own.