Hacker News Reader

To give the author the benefit of the doubt, maybe he plans to eventually make the computation happen asynchronously (e.g., in a web worker), or at least maybe he wants to leave that possibility open. Then the API wouldn't need to change.

NO it is not. Curryhoward is absolutely right: The reason behind this decision was to keep a consistent API when smartcrop might get broken into multiple calls to avoid blocking the mainloop for too long.

With that said, this should have been documented more clearly.

It's always easier to layer synchronous semantics on top of asynchronous than the other way around. Maybe the author intends on advanced usage using webworkers or a webservice.

If the callback were optional, the API call would've been completely fine IMHO.