Mirror of forum.nim-lang.org

223 :: Just a couple questions

• achesak (orginal) [2013-09-03T17:30:24+02:00] view original

  Hello again!

  As a way to help teach myself Nimrod, I've been working on porting a JavaScript math library that I wrote a while back. In doing this, however, I encountered a few problems.

  1. Several of the functions work with points in the Cartesian coordinate system, and return another point. Ideally I'd like for this to be as an array, but I can't figure out what to specify as the return type. I've tried something like:

  
  proc returnPoint(x : openarray[float], y : openarray[float]): openarray[float] =
      ...
      return [1.1,2.2]
  Run

  ...but this gives an error.

  2. Is there an equivalent to the integer mod operator for floats?

  Thanks in advance for any help!

---

• steved (orginal) [2013-09-03T18:29:33+02:00] view original

  AFAIK, openarray only works for function arguments, not return types. You have to return something like seq[float], and then you would need to say 'return @[1.1,2.2]' or use newSeq() and append values using add. Arrays are too inflexible to be useful return values, and openarray matches both arrays and sequences.

---

• Araq (orginal) [2013-09-03T20:49:30+02:00] view original

  You can also use a tuple[x, y: float] for a point.

---

• achesak (orginal) [2013-09-04T20:59:30+02:00] view original

  Thank you steved and Araq!

---

• gradha (orginal) [2013-09-10T19:55:23+02:00] view original

  1. Your example is confusing, are you accepting arrays of coordinates to join or just a single value? You can return arrays if you know the size in advance, otherwise you need to use sequences as mentioned:

  
  proc returnPoint(x, y: float): array[2, float] =
    result[0] = x
    result[1] = y
  
  proc returnPoints(x, y: openarray[float]): seq[array[2, float]] =
    assert len(x) == len(y)
    assert len(x) > 0
    result = newSeq[array[2, float]](x.len)
    for i in 0..high(x):
      result[i][0] = x[i]
      result[i][1] = y[i]
  
  echo "single ", repr(returnPoint(2.3, 4.5))
  echo "multi ", repr(returnPoints([1.0, 2.0, 3.0], [6.0, 7.0, 8.0]))
  Run

  2. Nimrod tends to wrap around the C library, so in C you would use fmod for this. A quick search in the nimrod index points to the math.fmod proc.

---