Mirror of forum.nim-lang.org

8767 :: `getTypeImpl` bug?

• geekrelief (orginal) [2022-01-03T21:48:33+01:00] view original

  https://play.nim-lang.org/#ix=3KTZ

  import macros
  
  type
    Vec[N: static[int], T] = object
      arr: array[N, T]
    Vec4[T] = Vec[4, T]
    Vec4f = Vec4[float32]
  
  var a: Vec4f
  
  macro test(t: typed) =
    echo t.treeRepr
    var res = t.getTypeImpl[1]
    echo res.treeRepr
    assert t == res
    
    echo t.getTypeImpl.treeRepr
    echo "\n"
    
    echo res.getTypeImpl.treeRepr
    echo "\n"
    
    assert t.getTypeImpl.repr == res.getTypeImpl.repr
  
  #test(a)
  test(Vec4f)
  Run

  I'm trying to understand the behavior of getTypeImpl https://nim-lang.github.io/Nim/macros.html#getTypeImpl%2Ctypedesc In the sample code above, t and res are equal, but what they return with another getTypeImpl is not the same. Is this a bug? I want what res.getTypeImpl returns.

---

• planetis (orginal) [2022-01-03T22:03:44+01:00] view original

  I don't think it's a bug. t is a typedesc[Vec4f] and that's what you see. Note the difference in with the macros entry. You are using a type but the docs use a variable.

---

• geekrelief (orginal) [2022-01-04T03:27:46+01:00] view original

  Ok, I think I get it. The confusing part is how t and res are both nnkSym Vec4f, but a getTypeImpl print out different things. It seems like res, since it comes out of a BracketExpr typedesc, carries some hidden info that it's actually a type.

---