Do we really want intermediate results?

[woutermont]

Now that we have separated these first general handlers, we can start to hone them to perfection.

A first question I have about them, is why their methods need the intermediate results.

• If their use case is to be able to safeHandle input, then I especially wonder why they are also included in the signatures of canHandleand handle.

• The case of safeHandle itself is then a question of whether safe handling should be implemented this way.

  • It is hardly any work for the calling code to write safe handling itself, given that they already possess the backup output. They can simply check the boolean output of canHandle and act accordingly:
    unsafeHandler.canHandle(input).pipe(mergeMap(handle => handle ? unsafeHandler.handle(input) : backup))
    No need to pass the backup through an extra method.

  • If we really want to take even that little bit of code out of the way, it can be interesting too see that the safeHandle method itself actually behaves like a Handler that takes another Handler as dependency and canHandle everything. A cleaner (i.e. more concern-separated) way of providing safe handling might therefore be to provide a SafeHandler wrapper: unsafeHandler.safeHandle(input, backup) would then be new SafeHandler(unsafeHandler, backup).handle(input).