lately i've become very careful not to pollute namespaces. this is sometimes hard to remember when using file and directory handles. normally, these are created inside the open statement, like open( FH, $file ) or die '...'; this is bad practice, because there are side-effects that some may not fully understand, and stricture rules don't apply to handles.

in the past, i've advocated the use of FileHandle to deal with scoping and stricture issues. i still use FileHandle, but i've come up with another way to avoid namespace pollution -- anonymous subroutines. it doesn't solve stricture problems, but i don't use strict in CGI production code anyway -- it speeds things up quite a bit. i do, and always will, use strict in development (please do the same!)

attached is a snippet for extracting files from a directory. notice the use of an anonymous subroutine, including passed parameters. i localize the handle within the sub, do my processing, and return an array -- and no messy namespace clutter to worry about later.

my $dir = '/path/to/files';

my @files = sub{
  local *DH;
  opendir( DH, $_[0] ) or die 'Cannot open ' . $_[0];
  return grep { -f $_ } readdir(DH);
}->( $dir );
[download]