We already await the promises that we set on the `encodedWith` instead of setting already
resolved promises to `encodedWith` we can set the resolved values
So, the users can use like
`const { mozjpeg: { binary } } = await image.encode({ mozjpeg })` or
they can first run
`await image.encode({ mozjpeg })` and then
`image.encodedWith.mozjpeg.binary`
Before this, there were one way to use the API:
call `await image.encode({ mozjpeg: {} })`
then use `encodedWith` by asserting that `mozjpeg` property exists on it
After adding return value to encode, people
will be able to use it like
`const { mozjpeg } = await image.encode({ mozjpeg: {} });`
which provides better type safety
Instead of returning `any` we're now returning the whole object.
Still from typing perspective, the API is not that
great since we don't have any type relation
between `encode` calls and `encodedWith` property. Maybe we can think about
returning directly from `encode` call
with the returned object having properties
that is supplied in `encode` calls
* Make decode module return value `ImageData`
* Fix global definition of ImageData
* Use concrete Encoder types for encode functions
* Use ArrayBufferView in FileLike instead of using a similar type
* Throw error when the `encode` functions
return null
* Use generic types for WorkerPool
* Fix `encode` function typing
in `index.ts`
* Remove ts-ignore for web-streams-polyfill
and handle nulls for TransformStream
* Fix rollup entry point (now we need to have
`index.ts` instead of `index.js`)
* Updated import to contain file extension `js`
* Separated WebAssembly definitions from missing-types
* Converted resizer.resize result to Uint8ClampedArray
* Moved ResizeInstantiateOptions to an interface
