When reading the unit test for the sub oneliner from t/oneliner.t

try_oneliner(q{$PATH = 'foo'; print $PATH},[], q{foo},   'dollar signs' );

You have the feeling that oneliner authorizes the usage of variables, all the most the test is passing.
But in truth it's using references:

# $(ABSPERLRUN)  -e '$$PATH = '\''foo'\''; print $$PATH' --

I wonder if we should teach sub oneliner to allow_variables when calling quote_literal, or detect incorrect usage of variable...

For example you can use %h but then it's going to fail when using $h later to access a value...

Detail:

my %h = (1..4); is a valid oneliner

$mm->oneliner( q[my %h = (1..4);]);
# $(ABSPERLRUN)  -e 'my %h = (1..4); print $$h[1]' --

but my %h = (1..4); print $h[1] is invalid

$mm->oneliner( q[my %h = (1..4);]);
# $(ABSPERLRUN)  -e 'my %h = (1..4); print $$h[1]' --

#!perl

use Test::More;

require ExtUtils::MM;

my $mm = MM->new(
	{ NAME => q[Fake], }
);

my $cmd;

# valid - using ref
my $cmd = $mm->oneliner( q[$PATH = 'foo'; print $PATH]);
note $cmd;
# $(ABSPERLRUN)  -e '$$PATH = '\''foo'\''; print $$PATH' --

# invalid
my $cmd = $mm->oneliner( q[my $PATH = 'foo'; print $PATH]);
note $cmd;
# $(ABSPERLRUN)  -e 'my $$PATH = '\''foo'\''; print $$PATH' --