URIs not Equivalent with Differently Ordered Query Strings

[oalders]

See https://rt.cpan.org/Ticket/Display.html?id=81491

I'm wondering if rather than

sub _obj_eq {
    return overload::StrVal($_[0]) eq overload::StrVal($_[1]);
}

we should do something like

sub _obj_eq {
    return $_[0]->canonical->as_string eq $_[1]->canonical->as_string;
}

I haven't tested this, but wanted to start a discussion here.

Ping @eserte and @theory who are commenters on the original RT.

[theory]

@srezic had a valid point in the original RT that order might be important in determining equivalence. He said:

I think order matters here. Somebody could use a query string like

foo.bar?ordered_array=1;ordered_array=2;ordered_array=3;ordered_array=4

and expect it to be ordered.

So I can imagine sometimes wanting it to be considered equivalent and sometimes not, depending on the use-case and context. Maybe it'd make sense to add a method that compares without regard to query string ordering?

[oalders]

Is there a use case where the order of parameters actually matters? (I haven't read the RFC). I'd be inclined to ensure the params were all ordered before doing the comparison.

[genio]

Yes, order matters. Mojo::Parameters's example should likely be followed.

11:19 <@sri> if you change order the query strings become semantically different according to the specs
11:19 <@sri> which may or may not be important, depending on the apps involved