Slick - Update full object or more than 22 columns

Solution 1:

Stefan Zeiger, the Slick lead, said we couldn't. He suggested however that we have nested projections over the flat 22+ columns table:

// 2 classes for the nested structure
case class Part(i1: Int, i2: Int, i3: Int, i4: Int, i5: Int, i6: Int)
case class Whole(id: Int, p1: Part, p2: Part, p3: Part, p4: Part)

// Note that it's a Table[Int] -- we only map the primary key in *
object T extends Table[Int]("t_wide") {
  def id = column[Int]("id", O.PrimaryKey)
  def p1i1 = column[Int]("p1i1")
  def p1i2 = column[Int]("p1i2")
  def p1i3 = column[Int]("p1i3")
  def p1i4 = column[Int]("p1i4")
  def p1i5 = column[Int]("p1i5")
  def p1i6 = column[Int]("p1i6")
  def p2i1 = column[Int]("p2i1")
  def p2i2 = column[Int]("p2i2")
  def p2i3 = column[Int]("p2i3")
  def p2i4 = column[Int]("p2i4")
  def p2i5 = column[Int]("p2i5")
  def p2i6 = column[Int]("p2i6")
  def p3i1 = column[Int]("p3i1")
  def p3i2 = column[Int]("p3i2")
  def p3i3 = column[Int]("p3i3")
  def p3i4 = column[Int]("p3i4")
  def p3i5 = column[Int]("p3i5")
  def p3i6 = column[Int]("p3i6")
  def p4i1 = column[Int]("p4i1")
  def p4i2 = column[Int]("p4i2")
  def p4i3 = column[Int]("p4i3")
  def p4i4 = column[Int]("p4i4")
  def p4i5 = column[Int]("p4i5")
  def p4i6 = column[Int]("p4i6")
  // This is just the default projection -- It doesn't have to contain all columns
  def * = id
  // Instead, we use nested tuples for a full projection:
  def all = (
    id,
    (p1i1, p1i2, p1i3, p1i4, p1i5, p1i6),
    (p2i1, p2i2, p2i3, p2i4, p2i5, p2i6),
    (p3i1, p3i2, p3i3, p3i4, p3i5, p3i6),
    (p4i1, p4i2, p4i3, p4i4, p4i5, p4i6)
  )
  // And override create_* to get the DDL for all columns.
  // Yeah, this is ugly. It used to be much simpler in ScalaQuery.
  // We can add a helper method to simplify it.
  override def create_* =
    all.shaped.packedNode.collect {
      case Select(Ref(IntrinsicSymbol(in)), f: FieldSymbol) if in == this => f
    }.toSeq.distinct
}

T.ddl.create
// Insert into T.all. The extra ".shaped" call is needed because we cannot
// get the types in an implicit conversion due to SI-3346
T.all.shaped.insert(
  0,
  (11, 12, 13, 14, 15, 16),
  (21, 22, 23, 24, 25, 26),
  (31, 32, 33, 34, 35, 36),
  (41, 42, 43, 44, 45, 46)
)

// Get the nested tuples in a query
val q1 = T.map(_.all)
println(q1.first)

// Map the result to the case classes
val i2 = q1.mapResult { case (id, p1, p2, p3, p4) =>
  Whole(id, Part.tupled.apply(p1), Part.tupled.apply(p2), Part.tupled.apply(p3), Part.tupled.apply(p4))
}
println(i2.first)

which is now a test at Slick including one for version 3. As for updating:

val oData = Whole(0,
  Part(11, 12, 13, 14, 15, 16),
  Part(21, 22, 23, 24, 25, 26),
  Part(31, 32, 33, 34, 35, 36),
  Part(41, 42, 43, 44, 45, 46)
)
val oData2 = Whole(10,
  Part(111, 12, 13, 14, 15, 16),
  Part(121, 22, 23, 24, 25, 26),
  Part(131, 32, 33, 34, 35, 36),
  Part(141, 42, 43, 44, 45, 46)
)

ts.ddl.create

ts.insert(oData)
assertEquals(oData, ts.first)

ts.filter(_.p1i2 === 12).update(oData2)
assertEquals(oData2, ts.first)

The nested objects with Slick's projections can be flattened for the single object that you bring in, or take away with.

Solution 2:

Some suggestions for problem 2:

• Could you use Slick 3.0? This version seems to have a solution
• Could you change the layout of the database in such a way permissions are rows instead of columns? This seems more extensible anyway