desc: A Promise is an object that represents a value that will exist in the future, but doesn't right now. Promises allow you to then attach callbacks that can run once the value becomes available (known as *resolving*), or if an error has occurred (known as *rejecting*).
Errors that occur during execution will be caught and turned into a rejection automatically. If `error()` is called with a table, that table will be the rejection value. Otherwise, string errors will be converted into `Promise.Error(Promise.Error.Kind.ExecutionError)` objects for tracking debug information.
* This should be used to abort any ongoing operations leading up to the promise being settled.
* Call the `onCancel` function with a function callback as its only argument to set a hook which will in turn be called when/if the promise is cancelled.
*`onCancel` returns `true` if the Promise was already cancelled when you called `onCancel`.
* Calling `onCancel` with no argument will not override a previously set cancellation hook, but it will still return `true` if the Promise is currently cancelled.
* You can set the cancellation hook at any time before resolving.
The same as [[Promise.new]], except execution begins after the next `Heartbeat` event.
This is a spiritual replacement for `spawn`, but it does not suffer from the same [issues](https://eryn.io/gist/3db84579866c099cdd5bb2ff37947cec) as `spawn`.
Begins a Promise chain, calling a function and returning a Promise resolving with its return value. If the function errors, the returned Promise will be rejected with the error. You can safely yield within the Promise.try callback.
::: tip
`Promise.try` is similar to [[Promise.promisify]], except the callback is invoked immediately instead of returning a new function.
:::
```lua
Promise.try(function()
return math.random(1, 2) == 1 and "ok" or error("Oh an error!")
Someone needs to consume this rejection (i.e. `:catch()` it), otherwise it will emit an unhandled Promise rejection warning on the next frame. Thus, you should not create and store rejected Promises for later use. Only create them on-demand as needed.
Accepts an array of Promises and returns a new Promise that resolves with an array of in-place PromiseStatuses when all input Promises have settled. This is equivalent to mapping `promise:finally` over the array of Promises.
If the first Promise to settle from the array settles with a rejection, the resulting Promise from `race` will reject.
If you instead want to tolerate rejections, and only care about at least one Promise resolving, you should use [[Promise.any]] or [[Promise.some]] instead.
Accepts an array of Promises and returns a Promise that is resolved as soon as `count` Promises are resolved from the input array. The resolved array values are in the order that the Promises resolved in. When this Promise resolves, all other pending Promises are cancelled if they have no other consumers.
`count` 0 results in an empty array. The resultant array will never have more than `count` elements.
Accepts an array of Promises and returns a Promise that is resolved as soon as *any* of the input Promises resolves. It will reject only if *all* input Promises reject. As soon as one Promises resolves, all other pending Promises are cancelled if they have no other consumers.
Resolves directly with the value of the first resolved Promise. This is essentially [[Promise.some]] with `1` count, except the Promise resolves with the value directly instead of an array with one element.
Returns a Promise that resolves after `seconds` seconds have passed. The Promise resolves with the actual amount of time that was waited.
This function is **not** a wrapper around `wait`. `Promise.delay` uses a custom scheduler which provides more accurate timing. As an optimization, cancelling this Promise instantly removes the task from the scheduler.
::: warning
Passing `NaN`, infinity, or a number less than 1/60 is equivalent to passing 1/60.
`Promise.each` is similar to `Promise.all`, except the Promises are ran in order instead of all at once.
But because Promises are eager, by the time they are created, they're already running. Thus, we need a way to defer creation of each Promise until a later time.
The predicate function exists as a way for us to operate on our data instead of creating a new closure for each Promise. If you would prefer, you can pass in an array of functions, and in the predicate, call the function and return its return value.
If the Promise a predicate returns rejects, the Promise from `Promise.each` is also rejected with the same value.
If the array of values contains a Promise, when we get to that point in the list, we wait for the Promise to resolve before calling the predicate with the value.
If a Promise in the array of values is already Rejected when `Promise.each` is called, `Promise.each` rejects with that value immediately (the predicate callback will never be called even once). If a Promise in the list is already Cancelled when `Promise.each` is called, `Promise.each` rejects with `Promise.Error(Promise.Error.Kind.AlreadyCancelled`). If a Promise in the array of values is Started at first, but later rejects, `Promise.each` will reject with that value and iteration will not continue once iteration encounters that value.
Returns a Promise containing an array of the returned/resolved values from the predicate for each item in the array of values.
If this Promise returned from `Promise.each` rejects or is cancelled for any reason, the following are true:
- Iteration will not continue.
- Any Promises within the array of values will now be cancelled if they have no other consumers.
- The Promise returned from the currently active predicate will be cancelled if it hasn't resolved yet.
params:
- name: list
type: "array<T|Promise<T>>"
- name: predicate
desc: The callback to call for each value in the list.
Within the failure handler, you should never assume that the rejection value is a string. Some rejections within the Promise library are represented by [[Error]] objects. If you want to treat it as a string for debugging, you should call `tostring` on it first.
Shorthand for `Promise:andThen(nil, failureHandler)`.
::: warning
Within the failure handler, you should never assume that the rejection value is a string. Some rejections within the Promise library are represented by [[Error]] objects. If you want to treat it as a string for debugging, you should call `tostring` on it first.
Similar to [[Promise.andThen]], except the return value is the same as the value passed to the handler. In other words, you can insert a `:tap` into a Promise chain without affecting the value that downstream Promises receive.
```lua
getTheValue()
:tap(print)
:andThen(function(theValue)
print("Got", theValue, "even though print returns nil!")
end)
```
If you return a Promise from the tap handler callback, its value will be discarded but `tap` will still wait until it resolves before passing the original value through.
If the Promise is cancelled, any Promises chained off of it with `andThen` won't run. Only Promises chained with `finally` or `done` will run in the case of cancellation.
Set a handler that will be called only if the Promise resolves or is cancelled. This method is similar to `finally`, except it doesn't catch rejections.
`done` should be reserved specifically when you want to perform some operation after the Promise is finished (like `finally`), but you don't want to consume rejections (like in <ahref="/roblox-lua-promise/lib/Examples.html#cancellable-animation-sequence">this example</a>). You should use `andThen` instead if you only care about the Resolved case.
Like `finally`, if the Promise is cancelled, any Promises chained off of it with `andThen` won't run. Only Promises chained with `done` and `finally` will run in the case of cancellation.
Promises are eager, so if you pass a Promise to `andThenReturn`, it will begin executing before `andThenReturn` is reached in the chain. Likewise, if you pass a Promise created from [[Promise.reject]] into `andThenReturn`, it's possible that this will trigger the unhandled rejection warning. If you need to return a Promise, it's usually best practice to use [[Promise.andThen]].
Returns a new Promise that resolves if the chained Promise resolves within `seconds` seconds, or rejects if execution time exceeds `seconds`. The chained Promise will be cancelled if the timeout is reached.
Rejects with `rejectionValue` if it is non-nil. If a `rejectionValue` is not given, it will reject with a `Promise.Error(Promise.Error.Kind.TimedOut)`. This can be checked with [[Error.isKind]].
Promises will only be cancelled if all of their consumers are also cancelled. This is to say that if you call `andThen` twice on the same promise, and you cancel only one of the child promises, it will not cancel the parent promise until the other child promise is also cancelled.
Chains a Promise from this one that is resolved if this Promise is already resolved, and rejected if it is not resolved at the time of calling `:now()`. This can be used to ensure your `andThen` handler occurs on the same frame as the root Promise execution.
If this Promise is still running, Rejected, or Cancelled, the Promise returned from `:now()` will reject with the `rejectionValue` if passed, otherwise with a `Promise.Error(Promise.Error.Kind.NotResolvedInTime)`. This can be checked with [[Error.isKind]].
Yields the current thread until the given Promise completes. Returns true if the Promise resolved, followed by the values that the promise resolved or rejected with.
::: warning
If the Promise gets cancelled, this function will return `false`, which is indistinguishable from a rejection. If you need to differentiate, you should use [[Promise.awaitStatus]] instead.
desc: Yields the current thread until the given Promise completes. Returns the Promise's status, followed by the values that the promise resolved or rejected with.
returns:
- type: PromiseStatus
desc: The Promise's status.
- type: ...any?
desc: The values that the Promise resolved or rejected with.
