type(x) != typeof(x): typeof(x) keeps the ntyAlias, unlike type(x)

[timotheecour]

after reviewing #16962 I tried to double check whether type(x) and typeof(x) have the same behavior (without the typeOfProc arg); it turns out it's not the case:

• Example 1

when true:
  block:
    type outType = type(0.0)
    echo outType
  block:
    type outType = typeof(0.0)
    echo outType

Current Output

float64
outType

Expected Output

both should behave the same, and both should print float64:
float64
float64

note that this already prints:

when true:
  block:
    echo type(0.0)
  block:
    echo typeof(0.0)

float64
float64

• Example 2

when true:
  import macros
  macro bar(T: typed) =
    let s = T.getTypeImpl[1]
    echo s.typeKind
  type outType1 = type(0.0)
  type outType2 = typeof(0.0)
  bar(outType1)
  bar(outType2)

prints:
ntyAlias
ntyFloat

this probably examples Example 1
I'd expect this to print:
ntyFloat
ntyFloat

• Example 3
  these should be refactored (in semmagic.nim, semtypes.nim):

proc semTypeof2(c: PContext; n: PNode; prev: PType): PType =
  openScope(c)
  var m = BiggestInt 1 # typeOfIter
  if n.len == 3:
    let mode = semConstExpr(c, n[2])
    if mode.kind != nkIntLit:
      localError(c.config, n.info, "typeof: cannot evaluate 'mode' parameter at compile-time")
    else:
      m = mode.intVal
  let t = semExprWithType(c, n[1], if m == 1: {efInTypeof} else: {})
  closeScope(c)
  fixupTypeOf(c, prev, t)
  result = t.typ

proc semTypeOf(c: PContext; n: PNode): PNode =
  var m = BiggestInt 1 # typeOfIter
  if n.len == 3:
    let mode = semConstExpr(c, n[2])
    if mode.kind != nkIntLit:
      localError(c.config, n.info, "typeof: cannot evaluate 'mode' parameter at compile-time")
    else:
      m = mode.intVal
  result = newNodeI(nkTypeOfExpr, n.info)
  let typExpr = semExprWithType(c, n[1], if m == 1: {efInTypeof} else: {})
  result.add typExpr
  result.typ = makeTypeDesc(c, typExpr.typ)

• Example 4
  => document typeof #16965
  the docs for typeof are bad; there should at least be a runnableExamples:
  

• Example 5
  this seems un-necessary in semMagic (and seems to pass test suite without it) which is a hot-spot and bypasses normal overload resolution:

  of mTypeOf:
    markUsed(c, n.info, s)
    result = semTypeOf(c, n)

(the other block like this in magicsAfterOverloadResolution seems correct though)

Additional Information

• 1.5.1 3eebbb2
• typeof was introduced in Regression: 0.18 code with mapIt() fails to compile on 0.19 #9093
• a similar bug as Example 1 was fixed in typeToString can now show (recursively) resolved type aliases; fixes #8569 #8083 #8570 #11678

[metagn]

both should behave the same, and both should print float64:
float64
float64

Is outType not supposed to be an alias in both of these situations? Is there supposed to be a special rule where if the RHS is an expression that resolves to a type, no alias is created?