Ok, I did some testing with PAR Packager, version 1.013 (PAR version 1.005)

 pp -M strict -M warnings -M Data::Dump -M Win32::Unicode::Native -M Win32::Process::CommandLine -M Win32::CommandLine   dragdropmeargv.pl

#!/usr/bin/perl --
use strict; use warnings;
use Data::Dump;
dd \@ARGV;
eval 'use Win32::Unicode::Native;';
dd \@ARGV;

{
    use Win32::Process::CommandLine;
    Win32::Process::CommandLine::GetPidCommandLine($$, my $str);
    dd $str;
}

use Win32::CommandLine();
dd [ Win32::CommandLine::argv() ];
dd [ Win32::CommandLine::command_line() ];
#~ scalar <>;
sleep 10;
[download]

But the results were disappointing, no unicode

[
  "utf-filedir\\kebab\\cevap.txt",
  "utf-filedir\\kebab\\ra\x9Enjic.txt",
  "utf-filedir\\kebab\\???????.txt",
  "utf-filedir\\kebab\\??????.txt",
  "utf-filedir\\kebab\\?????.txt",
  "utf-filedir\\kebab\\????.txt",
  "utf-filedir\\kebab\\ra\x9Enjic",
  "utf-filedir\\kebab\\cevap.txt",
]
[
  "utf-filedir\\kebab\\cevap.txt",
  "utf-filedir\\kebab\\ra\x9Enjic.txt",
  "utf-filedir\\kebab\\???????.txt",
  "utf-filedir\\kebab\\??????.txt",
  "utf-filedir\\kebab\\?????.txt",
  "utf-filedir\\kebab\\????.txt",
  "utf-filedir\\kebab\\ra\x9Enjic",
  "utf-filedir\\kebab\\cevap.txt",
]
"utf-filedir\\a.exe utf-filedir\\kebab\\cevap.txt utf-filedir\\kebab\\
+raznjic.txt utf-filedir\\kebab\\???????.txt utf-filedir\\kebab\\?????
+?.txt utf-filedir\\kebab\\?????.txt utf-filedir\\kebab\\????.txt utf-
+filedir\\kebab\\raznjic utf-filedir\\kebab\\cevap.txt"
[
  "utf-filedir\\kebab\\cevap.txt",
  "utf-filedir\\kebab\\ra\x9Enjic.txt",
  "utf-filedir/kebab/RANJI~1.TXT",
  "utf-filedir/kebab/0CA1~1.TXT",
  "utf-filedir/kebab/35AA~1.TXT",
  "utf-filedir/kebab/531D~1.TXT",
  "utf-filedir/kebab/DD0B~1.TXT",
  "utf-filedir/kebab/EVAP~1.TXT",
  "utf-filedir/kebab/cevap.txt",
  "utf-filedir\\kebab\\????.txt",
  "utf-filedir\\kebab\\ra\x9Enjic",
  "utf-filedir\\kebab\\cevap.txt",
]
[
  "utf-filedir\\a.exe utf-filedir\\kebab\\cevap.txt utf-filedir\\kebab
+\\ra\x9Enjic.txt utf-filedir\\kebab\\???????.txt utf-filedir\\kebab\\
+??????.txt utf-filedir\\kebab\\?????.txt utf-filedir\\kebab\\????.txt
+ utf-filedir\\kebab\\ra\x9Enjic utf-filedir\\kebab\\cevap.txt",
]
[download]

Inspired from your test result, I am really wondering if that's the encoding problem or if the argv actually turned into ANSI before pipe to my perl script...

I've wrote a little more in reply to a upper thread. http://www.perlmonks.org/?node_id=991005 Do you think this can mean something ?

Though, thank you very much for the help test! =)

Supposedly this is the same as using wmain, but works with mingw

# define UNICODE
# define _UNICODE
#include <windows.h>
#include <stdio.h>
#include <shellapi.h>

int __cdecl main()
{
   LPWSTR *szArglist;
   int nArgs;
   int i;
   int ix;
    int nLength;
    
    wprintf(L"commandline==\n%ws\n====\n\n\n", GetCommandLineW() );
   szArglist = CommandLineToArgvW(GetCommandLineW(), &nArgs);
   if( NULL == szArglist )
   {
      wprintf(L"CommandLineToArgvW failed\n");
      return 0;
   }
   else
   {
        for( i=0; i<nArgs; i++) printf("%d: %ws\n", i, szArglist[i]);
        for( i=0; i<nArgs; i++) wprintf(L"\n\n%d: %s\n", i, szArglist[
+i]);
//~ http://stackoverflow.com/questions/4014172/wprintf-format-type-spe
+cification-ws# c++ - wprintf format type specification %ws - Stack Ov
+erflow
//~ http://msdn.microsoft.com/en-us/library/ms647550%28VS.85%29.aspx# 
+ wsprintf function (Windows) 
        for( i=0; i<nArgs; i++) {
            nLength = wcslen( szArglist[i] );
//~             printf("\n{%d: nLength %d}\n", i, nLength );
            printf("\n$F[%d] = []; # nLength %d\n", i, nLength );
            for( ix=0; ix < nLength; ix++ ){
//~                 printf("%d: %d %#04X\n", i, ix, szArglist[i][ix] )
+;
//~                 printf("{%d: %d %#04X} ", i, ix, szArglist[i][ix] 
+);
                printf("$F[%d][%d]=%#04X; ", i, ix, szArglist[i][ix] )
+;
                if( 0 == (ix % 5) ) { printf("\n"); }
            }
        }
   }

// Free memory allocated for CommandLineToArgvW arguments.

   LocalFree(szArglist);
//~    sleep(20);
   printf("\n");
   system("PAUSE"); 
   return(1);
}
[download]

And I do appear to get some kind of unicode :)

$F[1] = []; # nLength 39
$F[1][24]=0X6B; $F[1][25]=0X65;

$F[1][26]=0X62; $F[1][27]=0X61; $F[1][28]=0X62; $F[1][29]=0X5C; $F[1][
+30]=0X107;

$F[1][31]=0X65; $F[1][32]=0X76; $F[1][33]=0X61; $F[1][34]=0X70; $F[1][
+35]=0X2E;

$F[1][36]=0X74; $F[1][37]=0X78; $F[1][38]=0X74;


$F[2][24]=0X6B; $F[2][25]=0X65;

$F[2][26]=0X62; $F[2][27]=0X61; $F[2][28]=0X62; $F[2][29]=0X5C; $F[2][
+30]=0X72;

$F[2][31]=0X61; $F[2][32]=0X17E; $F[2][33]=0X6E; $F[2][34]=0X6A; $F[2]
+[35]=0X69;

$F[2][36]=0X107; $F[2][37]=0X2E; $F[2][38]=0X74; $F[2][39]=0X78; $F[2]
+[40]=0X74;

binmode STDOUT, ':encoding(UTF-8)';
fudge(\@F);
use Data::Dump; dd\@F;
sub fudge {
    my( $args ) = @_;
    for my $arg ( @$args ){
        next unless defined $arg;
        my $one = "";
        for my $char ( @$arg ){
            next unless defined $char;
            my $chr = chr( $char );
            print $chr;
            $one .= $chr;
            
        }
        print "\n";
        dd $one;
    }
}

__END__
kebab\ćevap.txt
"kebab\\\x{107}evap.txt"
kebab\ražnjić.txt
"kebab\\ra\x{17E}nji\x{107}.txt"
[download]

Now all that is left is to figure out how to call system() in C

Win32::CommandLine doesn't actually use unicode calls, you need to compile it with

#define UNICODE
#define _UNICODE
[download]

Then it'll probably work

Although it should probably use GetCommandLineW explicitly