Handle async tasks in your svelte application with ease thanks to @sheepdog/svelte.
Note
Sheepdog was written and is maintained by Mainmatter and contributors. We offer consulting, training, and team augmentation for Svelte – check out our website to learn more!
@sheepdog/svelte supplies a simple way to introduce cancellable concurrency into your app. Not only do they provide the cancellability that is missing from normal Promises, @sheepdog/svelte also provides a public API that allows you to observe the running state of your task without having to set a single flag manually.
Tasks that live on components are automatically cancelled when their context is destroyed, meaning you don't need to worry about the clean up - we've got you covered.
Choose whether you want to keep the oldest, keep the newest or keep all instances of your task to help boost the performance of your app and reduce unnecessary server load.
Install it using your favorite package manager:
pnpm install @sheepdog/svelteThen put it to work immediately wherever you want cancellable promises. (Cancellation is only available when using the generator function or async transform, but more on that later).
All tasks will return a store with the same structure:
isRunning: Boolean - whether the task is currently running or notlast: TaskInstance | undefined - the last task instance, regardless of whether it errored, was canceled, or was successfullastCanceled: TaskInstance | undefined - the last canceled task instancelastErrored: TaskInstance | undefined - the last errored task instancelastRunning: TaskInstance | undefined - the last running task instance, as soon as the task stops running, this will be undefinedlastSuccessful: TaskInstance | undefined - the last successful task instanceperformCount: Number - the number of times the task has been run,
All TaskInstances will have the same structure:
error: undefined | unknown - if an error is thrown inside the task instance, it will be found hereisCanceled: boolean - whether the task instance was canceledisError: boolean - whether the task instance throw an error before completingisRunning: boolean - whether the task instance is currently runningisSuccessful: boolean - whether the task instance completed successfullyvalue: undefined | TReturn - if the task instance completed successfully, this will be the return value
There are several flavours of tasks to choose from (check out the interactive docs here [insert link to site when ready]).
With all types of task, it is possible to invoke it directly or add the kind parameter to the options object.
If you don't care about mid-call cancellation, then you can utilize the concurrency with any of the following task types.
If you do care about mid-call cancellation, be sure to check out the Task Cancellation section.
This simply gives you a task wrapper around your function. It will not handle any kind of concurrency for you.
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task(async (param: number) => {
await timeout(2000);
return param * 2;
});
</script>OR
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task(
async (param: number) => {
await timeout(2000);
return param * 2;
},
{ kind: 'default' },
);
</script>This will cancel the oldest instance of the task and start a new instance of it. You can also provide a max that will only restart the oldest task instance if the threshold is exceeded.
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task.restart(
async (param: number) => {
await timeout(2000);
return param * 2;
},
{ max: 3 },
);
</script>OR
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task(
async (param: number) => {
await timeout(2000);
return param * 2;
},
{ kind: 'restart', max: 3 },
);
</script>Both of the above will result in 3 simultaneous tasks being allowed to run. Triggering the task a fourth time will cancel the oldest task.
This will cancel any new instances of the task. You can also provide a max that will only drop the task instances if the threshold is exceeded.
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task.drop(
async (param: number) => {
await timeout(2000);
return param * 2;
},
{ max: 3 },
);
</script>OR
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task(
async (param: number) => {
await timeout(2000);
return param * 2;
},
{ kind: 'drop', max: 3 },
);
</script>Both of the above will result in 3 simultaneous tasks being allowed to run. Triggering the task a fourth time will cause it to be cancelled, leaving the initial 3 task instances untouched.
This will add all task instances to a list and each task will be run in order. You can also provide a max that will dictate the number of task instances that will run at the same time.
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task.enqueue(
async (param: number) => {
await timeout(2000);
return param * 2;
},
{ max: 3 },
);
</script>OR
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task(
async (param: number) => {
await timeout(2000);
return param * 2;
},
{ kind: 'enqueue', max: 3 },
);
</script>Both of the above will result in 3 simultaneous tasks being allowed to run. Any additional instances of the task will be added to a list and run when there is space in the queue.
This will run the initial tasks and then ensure that the very last task instance is also run. You can also provide a max that will dictate the number of task instances that will run initially. Note: keepLatest will only preserve the final one task instance.
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task.keepLatest(
async (param: number) => {
await timeout(2000);
return param * 2;
},
{ max: 3 },
);
</script>OR
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task(
async (param: number) => {
await timeout(2000);
return param * 2;
},
{ kind: 'keepLatest', max: 3 },
);
</script>Both of the above will result in 3 simultaneous tasks being allowed to run initially. Triggering the task again will wait for the oldest to complete and then run the latest task instance, preserving the most recent instance.
As the return value from the task wrapper is a store, you can access it just like you would with any other store (check the Task Instance section for more detail about what to expect):
<script>
import { task, timeout } from '@sheepdog/svelte';
const myTask = task(async (param: number) => {
await timeout(2000);
return param * 2;
});
</script>
{$myTask.isRunning}
{$myTask.last.value}
{$myTask.performCount}With normal Promises, once you have triggered it to run, the only way to interrupt it is to implement your own series of checks at different intervals. With @sheepdog/svelte we offer two ways to attain this ability without having to handle it yourself:
@sheepdog/svelte can utilize generator functions to give us fine-grain control of how far through our task will run when cancelled.
let data;
const instance = task(async function*() {
const res = async fetch('...');
yield;
data = await res.json()
});Using a generator function, we can now stop our function as soon as it is cancelled, that means that if we cancelled the task while the fetch request was running, the data attribute would not be reassigned.
@sheepdog/svelte also includes a vite plugin that transforms your async functions into generators.
To set this up, you simply need to important the vite plugin and add it to your plugins array:
// vite.config.ts
import { sveltekit } from '@sveltejs/kit/vite';
import { coverageConfigDefaults, defineConfig } from 'vitest/config';
+ import { asyncTransform } from '@sheepdog/svelte/vite';
export default defineConfig(({ mode }) => ({
- plugins: [sveltekit()],
+ plugins: [sveltekit(), asyncTransform()],
...
})
)Then you can use standard async functions in your task definition and under the hood, it will transform your code from this:
const instance = task(async () => {
const res = await fetch('...');
});to this:
const instance = task(async function* () {
const res = yield fetch('...');
});Meaning you get all of the functionality of generators without having to implement them yourself. And have no fear, this will only apply to async functions that you pass as a parameter to the task function from @sheepdog/svelte. (You can still use standard async promises as you would normally.)
As well as providing your own parameter as the first argument when creating your task function, you also have optional access to another object containing two useful utilities.
signal is the AbortSignal from the AbortController, this can be used to investigate the state of the signal of the current task instance.
link allows you to link the current task to another task, allowing automatic cancellation if the parent task is aborted, or if the child is cancelled, the parent will also be cancelled.
const parent_task = task(async () => {
const res = await fetch('...');
return res;
});
const child_task = task(async ({ my_param1, my_param2 }, { signal, link }) => {
const res = await link(parent_task).perform();
if (signal.aborted) {
console.log("I've been cancelled 😭");
}
});When a task is canceled, it will throw a CancelationError when it aborts the current controller. Though this can be useful, it's not something we ever want to show to the end user.
For this reason we expose didCancel which allows us to distinguish an actual error from one thrown internally by sheepdog.
import { didCancel } from '@sheepdog/svelte';
const parent_task = task(async () => {
const res = await fetch('...');
return res;
});
parent_task.perform().catch((e) => {
// return early if it's sheepdog cancelation error
if (didCancel(e)) return;
// do something with the actual error from our task
});In this example, our fetch call might throw an error or our task might be canceled. With didCancel we can check the error and ignore any cancelation errors, while doing something meaningful with any real errors that come from our task.
The timeout function provides a convenient way to wait for a certain amount of milliseconds. Its implementation comes down to a promisified setTimeout call.
import { task, timeout } from '@sheepdog/svelte';
const myTask = task(async () => {
await timeout(500);
const res = await fetch('...');
return res;
});If you want to write a new test for the async transformation you just need to create a code.js file in a new folder in ./src/lib/tests/expected-transforms. Try to give the folder a descriptive name and the run pnpm generate-expected. This will create a series of transform.js files in the various folder which will later be used to test the transform. If you are modifying the transform make sure to run the tests before running the generate-expected script!
P.s. If, after you run the script, you'll see a folder with code.js and no transform.js this means that in that case the transform will not apply.
@sheepdog/svelte is developed by and ©
Mainmatter GmbH and contributors. It is
released under the MIT License.