First, try not to use a delimiter for your regex that causes you to have to escape a lot more characters. More importantly, your regex has several mistakes. Trying to match "anything except this multi-character sequence" is almost always done wrong, even by top experts (at least a few times), so such isn't a big surprise. (:

For example, [^\/­][^\/]*? is really just the same as [^/­]+?, and so avoids matching single slashes rather than avoiding double slashes, as the construct hints at. So (?:(?:[^\/­][^\/]*?|)­*? boils down to ([^/]*?)*?, which is just an inefficient way of writing [^/]*?.

("|').*?\2 will stop matching too early for "This \"string\" with quotes" but can also "backtrack" and match too much. You really want to force this construct to only match exactly quoted strings. So, '([^'\\]+|\\.)*'|"([^"\\]+|\\.)*" instead.

Your "stuff I don't care about" needs to avoid matching quotes or slashes so that you don't just skip over a starting quote as "something I don't care about". So your regex needs something like ([^'"/]+|$quotes|...)*.

And a tricky part is the "skip over / but not over //". Something like (?<!/)/(?!/).

Which brings us to this:

$text =~ s<
  (^
    (?:
        [^/'"]+
      | '([^'\\]+|\\.)*'
      | "([^"\\]+|\\.)*"
      | (?<!/)/(?!/)
    )*
  )
  //.*
><$1>xgm;
[download]

Which likely has several bugs. Note that I didn't allow for \ to cause the comment to continue on to subsequent lines because I both believe and hope that such doesn't actually work in the languages that I use //-comments in.

Note that the pseudo tokenizer needs to match any constructs that could contain quotes or slashes so, for example, /* ... */ would need to be handled if such might be encountered.

Some interesting cases:

use warnings;
use strict;

while (<DATA>) {
    s/(
                (?:(?:[^\/][^\/]*?|)*?  # anything, except comment sym
+bol
                ("|').*?\2              # quoted string (and take ever
+ything)
                (?:[^\/][^\/]*?|)*?     # anything, except comment sym
+bol
              )*?
            )
            (?:
              \/\/                      # comment symbol
              [^\\]*                    # anything, except continuatio
+n
            )??$
          /$1/x;

    print;
}

__DATA__
/* //
*/
/*
*/
*str = "//\"//";
[download]

Prints:

/* */
/*
*/
*str = "//\"
[download]

use warnings;

use Regexp::Common qw /comment/;

while(<DATA>){
  my ($line, $simple, $RE, $tye) = ($_)x4;
  $simple =~ s#//[^\n]*$##s;
  $RE =~ s#$RE{comment}{Portia}#\n#;
  $tye =~ s#
  (^
    (?:
        [^/'"]+
      | '([^'\\]+|\\.)*'
      | "([^"\\]+|\\.)*"
      | (?<!/)/(?!/)
    )*
  )
  //.*
  #$1#xgm;
  print "  [DATA] $line";
  print "[simple] $simple";
  print "    [RE] $RE";
  print "   [tye] $tye";
  print "\n";
}
__DATA__
blah // comment
/* //
*/
/*
*/
*str = "//\"//";
[download]

Fails if '//' occurs in quoted text,

int foo = sprintf(
    line,
    "Comparing, //-comments in C/C++ act like #-comments in %s. ",
    "perl");
[download]

Update: oops, misread, see GrandFather's examples instead.