Async is everywhere
We could define -in a simplistic way- the async code as the code that, after calling it, returns an answer that is not immediately available and most of the time is executed in a background thread. For example, to connect to a BLE device or to download an image from server.
Nowadays async code is everywhere and tech companies and opensource communities have made efforts to make it more legible and easier to deal with. The solutions varies in name and format depending on the languages and the platforms: we can find simple callbacks, futures or promises, syntactic sugar like async-await or compiler tricks like coroutines.
Together with the solutions mentioned above, we have the Reactive Extensions framework. The Rx creates a channel of signals with all the future async inputs (Observable) that can be consumed by anyone who subscribe to it (Observer).
Besides, the signals inside the channel can be altered using well known functional operators like map, flatMap or reduce and the channels can be connected with other channels to create a Rx pipeline. For example, you can build a pipeline that waits for a user input that triggers a remote call to a rest API followed by a complex calculation over the result, all of it in few lines of code!
Applying the Rx style
If you’re a MacOs or iOS developer is very common to find system APIs that needs delegates to deal with the async calls: CBCentralManager, CBPeripheral or NFCReaderSession for example. Although they could break our app style introducing another way to deal with asynchronism, move them to the bright side can be accomplished in two easy steps:
- Create a utility class that mimic the API interface.
- Add the delegate as internal class to intercept the answers and transforms it into observables.
Let’s create it for the
Once the utility class is ready, we can use the API in a more Rx way:
Understanding and predictability is very important in the context of an application development and in many cases is better to introduce a bit more code but with a high impact in the whole codebase readability.
If you are already using Rx libs, maybe the introduction of a Rx layer on top of a system API takes some effort but at the moment of use it together with other APIs or with UI code they will fit lot better, increasing productivity and flexibility.