I'm trying to create an object that will fetch a resource from the web, and needs to remember both where the resource was found eventually, as well as what the original URL was that we gave it.
I don't want to have to specify the URL twice, nor do I want to have loads of conditionals every time I want to use the URL to figure out whether I should use the "url" attribute or the "updated URL" attribute or some such, so I thought I'd create a second attribute with default property set to initialize from the original URL:
package foo;
use Moose;
has 'url' => (
is => 'rw',
isa => 'Str',
required => 1,
);
has 'original_url' => (
is => 'ro',
isa => 'Str',
default => sub { shift->url },
);
If this would have worked (see below), then I would have been able to change the url at will, and just use the url attribute whenever I need to access the "current" URL; and then later on, if I need to see if the url was ever changed, I can just compare the url attribute against the original_url attribute, and go from there.
The problem with this, however, is that the order of initialization of these attributes seems to be not well defined; sometimes the default sub for the original_url attribute is called before the url property has a value, which means it will return undef, which obviously causes an error.
I thought of making original_url a lazy attribute, add a predicate and then add a trigger on url to update original_url to the old value if its predicate says it hasn't been set yet, but it turns out that triggers are called after the value has been changed, so I don't think I can do that.
Is there a method to accomplish what I'm trying to do that won't cause errors?