Mirror of forum.nim-lang.org

5441 :: template/generic instantiation of `map` warning

• pmags (orginal) [2019-11-02T20:45:46+01:00] view original

  I'm working with derived enum types and I'm encountering errors "template/generic instantiation of map" warnings that I don't understand.

  Here's a shortened version of my code:

  import sequtils, tables
  
  type
    AmbigNucleotide = enum
      T, A, G, C, U, W, S, M, K, R, Y, B, D, H, V, N
    DNA = range[T..C]
    RNA = range[A..U]
  
  const
    tblDNAtoRNA = {A: A, T: U, G: G, C: C}.toTable
  
  proc transcribe(n: DNA): RNA =
    RNA tblDNAtoRNA[n]
  
  proc transcribe(ns: seq[DNA]): seq[RNA] =
    map(ns, transcribe)
    # I've also tried map[DNA,RNA](ns, transcribe) with the same results
  
  let dna = @[DNA A, T, T, G, C, C, T, T]
  let rna = transcribe(dna)
  
  echo "DNA: ", dna
  echo "Transcribed RNA: ", rna
  Run

  The code above compiles and works but I get the following warnings (edited to highlight key lines):

  
  example02.nim(17, 6) template/generic instantiation of `map` from here
  sequtils.nim(311, 10) Warning: Cannot prove that 'result' is initialized. This will become a compile time error in the future. [ProveInit]
  Run

  My key point of confusion is why this is a generic instantiation of map at all. It seems the types are well defined.

---

• juancarlospaco (orginal) [2019-11-02T21:24:20+01:00] view original

  template/generic instantiation of FOO from here just means that the problem was inside a template or generic (the problem itself is not that line).

  For the error in this specific case check the documentation of Table, it should be initialized, thats the reason for the warning.

---

• pmags (orginal) [2019-11-02T23:40:50+01:00] view original

  Isn't tblDNAtoRNA already initialized by to.Table in the const declaration?

---

• mratsim (orginal) [2019-11-03T00:20:54+01:00] view original

  What's your Nim version?

  Your code compile as is for me on both Nim 0.20.2 and Nim 1.0.2.

---

• pmags (orginal) [2019-11-03T00:23:37+01:00] view original

  
  $ nim --version
  Nim Compiler Version 1.0.2 [MacOSX: amd64]
  Compiled at 2019-10-22
  Copyright (c) 2006-2019 by Andreas Rumpf
  
  git hash: 193b3c66bbeffafaebff166d24b9866f1eaaac0e
  active boot switches: -d:release
  Run

  As I noted, it compiles (and gives the right output) but gives the warning I described above.

---

• kobi (orginal) [2019-11-03T01:05:34+01:00] view original

  it's not an error, but at least in vscode it shows up as one. (in regular compilation, the text is not prefixed with a hint, maybe that's why it's taken as an error? a little annoying but doesn't affect your code. it will compile.

---

• juancarlospaco (orginal) [2019-11-03T01:30:21+01:00] view original

  Its a Warning not an Error. Should compile.

---

• jasper (orginal) [2019-11-03T02:10:07+01:00] view original

  You get this warning because 0 (or the enum equivalent) is not a valid value for the type RNA and newSeq elements are initialized to 0. In the case of map the warning is wrong since the elements are properly initialized but the compiler couldn't follow the control flow.

  Example:

  proc test =
    var x = newSeq[range[1..1]](5)
  Run

  You can add --warning[ProveInit]:off to your nim.cfg or add {.push warning[ProveInit]: off.} to your file.

  Unrelated tip: If a table is constant and the keys are contiguous, you can just use an array to avoid hashing/lookups.

  const
    tblDNAtoRNA = [T: RNA U, A: A, G: G, C: C]
  
  proc transcribe(n: DNA): RNA =
    tblDNAtoRNA[n]
  Run

---