I wrote this for the new forum, but it won’t let me post, so I’ll put it here: Guyren Howe guyren@relevantlogic.com frest.substack.com @unclouded lydb.xyz My major thesis for the future of coding is that SQL is a gaping wound in the computer industry. The relational model is a straightforward way to use declarative First Order Logic to retrieve and manipulate data. But SQL is a horrid way to achieve this. Datalog (https://en.wikipedia.org/wiki/Datalog#:~:text=Datalog%20is%20a%20declarative%20logic,behavior%20and%20properties%20from%20Prolog.), which is basically the pattern matching that all the languages are acquiring now, but for data, is the natural alternative. In fact, most of most business software (heck, most of most software of any kind) can in fact be expressed through the relational model. The only parts of typical software that needs to be written in a Turing Complete language are: • side-effecting things (e.g. sending email) • efficiency-focused computation (e.g. signal processing) • some types of complex sequential logic (e.g. some aspects of UIs) Everything else can and should be expressed in Datalog. The advantage of so doing for developers would be software that is more robust, simple and flexible. The advantage for non-developers is that e.g. FileMaker demonstrates that non-developers find manipulating relations natural, so this is also a way to let non-programmers do computing more easily. I’ve developed some theory around what software like that would look like. I discuss this at frest.substack.com. Maybe start with https://frest.substack.com/p/what-is-frest

Just to stave off any confusion, especially for new members just joining — there is no "new forum" yet. @Taylor Troesh built a thing and generously offered it up as something we could look at while we're considering what to do to move off of slack. The thing he built is not going to become the new official thing (we have different goals), but it will help us by giving us something to evaluate and reflect on. So for the time being, please keep using this slack.

And welcome to the club. Your pro-Datalog rant has been received, cataloged, and approved. 😉

I like the sound of a lot of your statements.

If ordinary folk can do functional programming, and professional programming languages are all turning into LISP/Scheme/ML, then FREST will too.

This one particularly resonated. One thing I'm not entirely sure about, What is Frest? I assumed a language until I read.

Work is underway on example implementations in more than one language.

I'm about 3 years into my own experiments, a lot of them around end users can use real languages. so I am keen to see where this goes

Frest is a design philosophy. It will most immediately produce a set of protocols for relational APIs with UI generation.