Mirror of forum.nim-lang.org

859 :: Wrong file names created in Windows

• AxBen (orginal) [2015-02-12T14:07:15+01:00] view original

  On Windows 7, the following programs (both saved as Latin-1 source) create:

  • C (gcc) => "tatü2.txt" [ü = \x00\xFC]
  • Nim (0.10.3 latest) => "tat�2.txt" [ü = \xFF\xFD]

  The file names appear as shown above in Windows's Explorer (UTF-16). Unless this is intentional, I'd consider this a bug...

  C:

  #include <stdio.h>
  void main() {
     FILE *h = fopen("tatü2.txt", "w+");
     fclose(h);
  }
  Run

  Nim:

  import os
  var f = open("tatü2.txt", fmAppend)
  close(f)
  Run

---

• BlaXpirit (orginal) [2015-02-12T14:12:41+01:00] view original

  All Nim source files are in the UTF-8 encoding (or its ASCII subset). Other encodings are not supported.

  Lexical analysis - Nim Manual

---

• AxBen (orginal) [2015-02-12T14:25:39+01:00] view original

  @BlaXpirit: Guessed you would say that ;-)

  Saved the Nim source as UTF-8 (as I always do, even on Windows). Now the file name is correct. But try this for the created file:

  for kind, path in walkDir("."):
     echo repr path
     try:
        moveFile(path, "ABC")
     except:
        echo "nope"
  
  try:
     moveFile("tatü2.txt", "DEF")
  except:
     echo "nope again"
  Run

  which gives me:

  000000000043F050".\\tat\195\1882.txt"
  nope
  nope again

---

• oswjk (orginal) [2015-02-12T21:35:19+01:00] view original

  moveFile() seems to use C's rename() function. I'm not sure that's going to use any UTF-8 filenames. I think it should preferably use MoveFileW() or _wrename() to support unicode.

---

• Araq (orginal) [2015-02-13T02:04:50+01:00] view original

  Thanks. Fixed.

---

• AxBen (orginal) [2015-02-13T08:41:02+01:00] view original

  Araq: Thx for fixing this!

---