A quick CPAN search suggests perhaps File::Wildcard might suit you, though its syntax is a bit strange. Personally, I'd probably just use File::Find::Rule.

Thanks. I think I might go with File::Wildcard. Its syntax is a bit strange, but it's not entirely baroque.

The problem with File::Find::Rule, for my purposes, is that I want to be able to specify the search rule entirely in one succinct string of text, such as a user might supply as a commandline argument. The power of File::Find::Rule is largely in its API, AFAICT.

Update: I have integrated File::Wildcard into my program over here, and it works like a champ.

A word spoken in Mind will reach its own level, in the objective world, by its own weight

It shouldn't be terribly hard to implement that yourself using File::Find or File::Find::Rule, just a bit of parsing of the path string. Not much code and it will work exactly how you want.

                - Ant
                - Some of my best work - (1 2 3)

Perhaps open a pipe to zsh and let it do the globular lifting?

## Off the cuff . . .
sub zglob {
  my $glob = shift;

  open( my $z, qq{zsh -c 'print -N "$glob"' |} ) or die "can't open gl
+obpipe: #!\n";

  local( $/ ) = "\0";

  my @ret;
  while( <$z> ) {
    chomp;
    push @ret, $_;
  }

  close( $z );

  return @ret;
}
[download]

Hmmm... maybe combine Class::XPath with Path::Class.

One could of course read in the whole filesystem below the toplevel directory you intend to use, turn it into an XML-structure and then apply XPath rules to it.

Something says me it will not be very efficient ...

CountZero

"If you have four groups working on a compiler, you'll get a 4-pass compiler." - Conway's Law

Something says me it will not be very efficient ...

The particular implementation you mentioned, serializing the file system tree into an XML doc then deserializing it into a XML tree, would indeed be inefficient.

However, XPath can be applied to tree structures (such as a file system) in an efficient manner. It was designed to do just that.