In general, meta-quoting for regexes (as used in split) can be done with quotemeta or with the  \Q ... \E intepolation escape sequences (see Quote and Quote-like Operators), which work for both double-quote and regex interpolation.

Are you sure this works with File::Glob 's meta characters sufficiently well?

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery FootballPerl is like chess, only without the dice}

No. In fact, I hadn't even thought about the general case in which glob metacharacters may be included in the input string, i.e., the string to be permuted (if that's what you're referring to).

Ferinstance, consider the input string 'Prefix{u,v}2={w,x}:b,{y,z}:1,2'. An attempt to fix this up so that glob produces

Prefix{u,v}2={w,x}:b:1
Prefix{u,v}2={w,x}:{y,z}:1
Prefix{u,v}2={w,x}:b:2
Prefix{u,v}2={w,x}:{y,z}:2
[download]

shows that there's a problem with the  , metacharacter: I can't figure out how to metaquote it.

c:\@Work\Perl\monks\NorthernFox>perl -wMstrict -MData::Dump -le
"my @proto_globules = (
   'Prefix{u,v}2={w,x}:b,{y,z}:1,2',
   );
 ;;
 my $rx_comma_in_curlies = qr{ { .*? , .*? } }xms;
 ;;
 my $rx_globmeta = qr{ [\[\]{}*?~] }xms;
 ;;
 my %globize = ('=' => '={',  ':' => '}:{',  '' => '}');
 ;;
 for my $globule (@proto_globules) {
   print qq{A: '$globule'};
   $globule =~ s{ ($rx_comma_in_curlies) }
                { (my $r = $1) =~ s{,}{\\,}xmsg;  $r; }xmsge;
   $globule =~ s{ (?= $rx_globmeta) }{\\}xmsg;
   print qq{B: '$globule'};
   $globule =~ s{ ([=:] | \z) }
                { print qq{S: '$1'};  $globize{$1}; }xmsge;
   print qq{C: '$globule'};
   my @globs = glob $globule;
   dd 'DD:', \@globs;  print '';  next;
   }
"
A: 'Prefix{u,v}2={w,x}:b,{y,z}:1,2'
B: 'Prefix\{u\,v\}2=\{w\,x\}:b,\{y\,z\}:1,2'
S: '='
S: ':'
S: ':'
S: ''
C: 'Prefix\{u\,v\}2={\{w\,x\}}:{b,\{y\,z\}}:{1,2}'
(
  "DD:",
  [
    "Prefix{u\\,v}2={w\\:b:1",
    "Prefix{u\\,v}2={w\\:b:2",
    "Prefix{u\\,v}2={w\\:{y\\:1",
    "Prefix{u\\,v}2={w\\:{y\\:2",
    "Prefix{u\\,v}2={w\\:z}:1",
    "Prefix{u\\,v}2={w\\:z}:2",
    "Prefix{u\\,v}2=x}:b:1",
    "Prefix{u\\,v}2=x}:b:2",
    "Prefix{u\\,v}2=x}:{y\\:1",
    "Prefix{u\\,v}2=x}:{y\\:2",
    "Prefix{u\\,v}2=x}:z}:1",
    "Prefix{u\\,v}2=x}:z}:2",
  ],
)
[download]

The  \{u\,v\} isn't too bad, but wrapping a pair of  { } permuting curlies around  \{w\,x\} or  b,\{y\,z\} reveals that the comma is still "active":

    still active! -------+------------+
                         |            |
                         v            v
C: 'Prefix\{u\,v\}2={\{w\,x\}}:{b,\{y\,z\}}:{1,2}'
                    ^        ^ ^          ^
                    |        | |          |
 in added curlies --+--------+-+----------+
[download]

So unless someone can figure out how to metaquote the glob comma-in-curlies operator, it looks like a parser of some kind is the way to go in the general case.

---

Give a man a fish:  <%-{-{-{-<

But in the case that I'd worry about such input I'd rather write a loop multiplying arrays.

That's easier to test. :)

Yep. Use a tool that already handles that 😏

Which tool exactly fulfills the OP's requirements? Could you demonstrate?
- Ron