Hacker News Reader

Comment by erezsh | Hacker News Reader

I'm trying to replace SQL by building a language that compiles to SQL. It has first-class functions, nicer syntax, better type-system, introspection, and other things you would expect from a modern language. But in the end, you still get SQL's performance, and the ability to use it with dozens of database engines.

View on HN

Have you seen the R library dplyr? It is a weirdly powerful data analysis tool, that can also produce SQL queries!

I heard about it but didn't know it can produce SQL. Thanks for the tip!

How is this different than a heavy-weight ORM like entity framework?

What I understood is that this like what TypeScript is to JavaScript (or to be more accurate Haxe to the few languages it transpiles to). It's a new language that will transpile to different SQL syntaxes instead of the method chaining approach of your typical ORM (that'll likely also include model abstractions over your tables and maybe also migration managing).

yea, I read it the same way, but ORM's like EF let you leverage an existing language to do basically the same thing as this does. Abstraction, data types, etc... I am unsure as to the value of something like that unless it is included with a client application that handles the conversion natively.

Oh man, I've dreamed about something along these lines too. I'd kill for algebraic data types, including option types instead of weird ternary logic for nulls.

Me too. I'm interested in what you've done. Please share a link if you have anything published.

My project is here: https://docs.racket-lang.org/plisqin/index.html I'm days away from a Show HN at this point. If I can pick only one idea that everyone should copy, it's that joins are values, not language constructs.

I don't know if it's exactly what you describe but Hasura compiles GraphQL queries and mutations to SQL (for Postgrs).

Yes, it's a lot like Hasura, but it's a full-fledged query language. GraphQL is a little limited in expression.

If this can allow me to write complex SQL (lots of JOINs, Multiple CTEs building on top of each other, Analytic functions etc), then I'm definitely interested. I've created an SQL generation engine in the past and can definitely appreciate how hard the problem is.

It already can. I'm a few weeks away from releasing a private alpha, so if you're interested I'll be happy to let you try it in exchange for feedback.

I'd love to. Hit me at me@manishgill.com :)

You mean you work on LINQ?

Only half joking here.

LINQ did a tiny step in the right direction. I'm aiming a lot further.

Is there a Clojure datalog to SQL compiler out there?

There was a ClojuRU 2019 talk about such a thing: https://youtu.be/Hf57HbRvYhM

As someone who spent handreds of hours writing (and cursing) SQL queries -it definitely sounds compelling.

Just a tiny teaser, you could write something like

    Person { country => name }

And it will output the following SQL:

    -- target: postgres
    SELECT country, array_agg(name) AS name FROM Person GROUP BY country

Is this not the same as graphql?

Sort of, GraphQL is just an interface that describes your data models, but you need to implement "resolvers" for each operation that tells the server how to fetch/mutate the queried data.

So you might have a GQL type like:

    type User {
      name: String
      age: Int
    }

But you need to implement a query with business logic for fetching this data, and that can be anything (but generally, a DB call).

Projects like Hasura kind of blur this line, as it's essentially one big GQL -> SQL JIT compiler in Haskell though.

Reply on news.ycombinator.com