Rafal Dittwald, “Data Oriented Programming”

The term “data-oriented programming” is often mentioned in the digital halls of the Clojure community, but what does it mean? In this talk, Rafal will attempt to distill the zeitgeist to explain the relatively new paradigm and answer: Is it any good? Is it just a fad? Can it be applied outside of Clojure?

Bio: Rafal is an entrepreneur and software engineer from Toronto. He used to work with PHP, Ruby and Javascript, but for the last ten years, he’s been building web apps exclusively in Clojure and Clojurescript. He occasionally comes out of his cave to talk about programming.

[marksto]

This static typing aspect was nicely addressed in Rich Hickey's "Maybe Not" talk which as well covered some conceptual issues in the Clojure's own spec library.
That said, this is not about statically typed systems nor about specs — this is all about picking up concepts more accurately whatever the system one is building.
For instance, if you heavily lean onto things like Optional/Maybe/Either because of lacking proper Union types in your language's type system, you are screwed.

[bgaRevoker]

When will you make a book, Rafal Dittwald ? So much software wisdom in your way of thinking and talking.

[DrewIsFail]

I think it's really important to prompt people to be very specific and give an example of how the static types are going to help them refactor. I think most examples people will come up with will revel that the equivalent problem will be modeled in clojure in such a way that the code won't need to change in most places and in places where it does the change will be easy to identify.

E.g if i remove a key from a hashmap all places where that key are used explicitly will be easy to find via a text search or using LSP.

I feel a lot of type issues arise from a lack of a good graph query language and a way to transfer data around a system.

[TankorSmash]

Well now I'm just curious about what happened after the end of the video!