Does it really make sense for ClassName to accept objects overloading stringification?

[tobyink]

Consider:

has ua_class => (
  is       => 'ro',
  isa      => t('ClassName'),
  default  => 'HTTP::Tiny',
);

has ua => (
  is       => 'ro',
  isa      => t('Object'),
  default  => sub { shift->ua_class->new() },
);

If somebody sets ua_class to an object overloading stringification, and the stringification yields a valid classname, then what will ua be?

This seems a source for very hard to diagnose bugs, especially as things like warn $self->ua_class will hide the issue.

[autarch]

This is a good question, but I don't have a good answer. I feel like I could make an argument either way.