Molecule vs other query languages

Every database language has its strengths and weaknesses and we’ll try here to compare existing languages with Molecule and see how they accomplish similar tasks.

Many query languages let you define data objects matching your domain that are then consumed by the host language (Scala) in combination with DSL tokens. Molecule instead let you use your domain terms directly as the query tokens.


If we take Slick for instance we could say that it “consumes” the domain terms coffees and suppliers in this Scala sequence comprehension:

// Slick
val coffeeSupliers = for {
  c <- coffees
  s <- suppliers
} yield (,

Whereas in Molecule, we only declare which attributes we are interested in. Molecule attributes themselves form the query - they are not consumed by an outer construct. That way, the domain terms directly form the query without additional keywords and constructs. Only the final get is a query keyword:

// Molecule
val coffeeSupliers =

We get the exact same type-inferred result back, a Future[List[(String, String)]]


Squeryl also consumes data objects, now in another DSL construct:

// Squeryl
def songs = from(MusicDb.songs)(s => where(s.artistId === id) select(s))

In Molecule we don’t need to use keywords like from, where and select (apart from the final get method). We instead get song entities (e) filtered by Artist by simply applying a required value directly to the artist attribute:

// Molecule
val songs = Song.e.artist_(id).get

(The underscore after artist makes the attribute tacit - that we don’t need to return its value)


Let’s compare with Datomic / Datalog