How to filter and update a table in Slick using a composite key?

Question

I have a table:

trait Schema {
    val db: Database

    class CoffeeTable(tag: Tag) extends Table[Coffee](tag, "coffees") {
        def from = column[String]("from")
        def kind = column[String]("kind")
        def sold = column[Boolean]("sold")
    }

    protected val coffees = TableQuery[Coffee]
}

I want to update records that are for sale. Here is my method:

  def markAsSold(soldCoffees: Seq[Coffee]): Future[Int] = {
    val cmd = coffees
      .filter { coffee =>
        soldCoffees
          .map(sc => coffee.from === sc.from && coffee.kind === sc.kind)
          .reduceLeftOption(_ || _)
          .getOrElse(LiteralColumn(false))
      }
      .map(coffee => coffee.sold)
      .update(true)
    db.db.stream(cmd)
  }

While it works for a small collection soldCoffee, it fails badly when entering hundreds of elements:

java.lang.StackOverflowError
    at slick.ast.TypeUtil$$colon$at$.unapply(Type.scala:325)
    at slick.jdbc.JdbcStatementBuilderComponent$QueryBuilder.expr(JdbcStatementBuilderComponent.scala:311)
    at slick.jdbc.H2Profile$QueryBuilder.expr(H2Profile.scala:99)
    at slick.jdbc.JdbcStatementBuilderComponent$QueryBuilder.$anonfun$expr$8(JdbcStatementBuilderComponent.scala:381)
    at slick.jdbc.JdbcStatementBuilderComponent$QueryBuilder.$anonfun$expr$8$adapted(JdbcStatementBuilderComponent.scala:381)
    at slick.util.SQLBuilder.sep(SQLBuilder.scala:31)
    at slick.jdbc.JdbcStatementBuilderComponent$QueryBuilder.expr(JdbcStatementBuilderComponent.scala:381)
    at slick.jdbc.H2Profile$QueryBuilder.expr(H2Profile.scala:99)
    at slick.jdbc.JdbcStatementBuilderComponent$QueryBuilder.$anonfun$expr$8(JdbcStatementBuilderComponent.scala:381)
    at slick.jdbc.JdbcStatementBuilderComponent$QueryBuilder.$anonfun$expr$8$adapted(JdbcStatementBuilderComponent.scala:381)
    at slick.util.SQLBuilder.sep(SQLBuilder.scala:31)

So the question is, is there any other way to do such an update?

What comes to my mind is to use easier SQL-code or the introduction of some artificial column that contains the concatenated values of columns fromand typeand filtering against him.