my $self = shift;
my $metaclass = $self->associated_metaclass;
+ my $expected_class = $self->_expected_constructor_class;
+
# If any of our parents have been made immutable, we are okay to
# inline our own new method. The assumption is that an inlined new
# method provided by a parent does not actually get used by
for my $meta (
grep { $_->is_immutable }
map { ( ref $metaclass )->initialize($_) }
+ grep { $_ ne $expected_class }
$metaclass->linearized_isa
) {
my $transformer = $meta->get_immutable_transformer;
+ # This is actually a false positive if we're in a subclass of
+ # this class, _and_ the expected class is not overridden (but
+ # should be), and the real expected class is actually
+ # immutable itself (see Fey::Object::Table for an example of
+ # how this can happen). I'm not sure how to actually handle
+ # that case, since it's effectively a bug in the subclass (for
+ # not overriding _expected_constructor_class).
return 1 if $transformer->inlined_constructor;
}
if ( my $constructor = $metaclass->find_method_by_name( $self->name ) ) {
my $class = $self->associated_metaclass->name;
- my $expected_class = $self->_expected_constructor_class;
if ( $constructor->body != $expected_class->can('new') ) {
warn "Not inlining a constructor for $class since it is not"