Thanks. It works! But I want to understand why it works. Can you explain what is happening in this line? *{'update_' . $n }->(); download I thought glob is only used for directories.

The glob() function is used to search for files.

On the other hand, perl has many different types: scalars, arrays, hashes, etc. And a perl program can happily have variables named $x, @x, and %x, and it will run without error. A typeglob is one of the types in perl. The typeglob *x represents all variables named 'x' in your program, e.g. $x, @x, %x, sub x {}, etc. In order to understand typeglobs, you need to read perlreftut first, so that you know what references are.

The code fragment:

*{'update_' . $n }
[download]

resolves to the typeglob: *update_test. The arrow following the brackets:

*{'update_' . $n }->
[download]

...dereferences the reference on the left of the arrow. The type of dereference is determined by what's on the right of the arrow:

*{'update_' . $n }->()
[download]

In this case, the left side of the arrow, i.e. the typeglob, is treated as a reference to a subroutine, and the subroutine is called. Compare to this:

#####use strict;
use warnings;
use 5.010;

sub greet {
    my $greeting = shift;
    say $greeting;
}

%greet = (
    a => 'hello',
    b => 'goodbye',
);


my $code_ref = \&greet;
$code_ref->('hello');

my $hash_ref = \%greet;
say $hash_ref->{b};

*greet->('goodbye');
say *greet->{a};


--output:--
hello
goodbye
goodbye
hello
[download]