Add a no path case to urls about rust-url HOT 9 CLOSED

As far as I understand, https://url.spec.whatwg.org/#urls makes no difference between an empty path and no path. Maybe this is a spec issue?

CC @annevk

from rust-url.

That is correct. In most cases it is not necessarry to make a difference between an empty path and no path. Since the url spec is browser-centric it probably decided not to support no paths.

from rust-url.

I searched for no-path in the referenced document and there were no hits.

from rust-url.

The empty vector represents the path /.

It doesn't seem like that.

extern crate url;

use url::Url;

fn main() {
    let a = Url::parse("http://example.com:80").unwrap();
    let b = Url::parse("http://example.com:80/").unwrap();
    println!("{:?}", a.path());  // Some([""])
    println!("{:?}", b.path());  // Some([""])
}

Perhaps one could use the empty vector for "no path" then?

from rust-url.

Yes one could do this :) This would not break (much) existing code.

from rust-url.

@pyfisch I don’t see in the HTTP RFC what defines that OPTIONS http://www.example.org:8001 and OPTIONS http://www.example.org:8001/ are different. What am I missing?

from rust-url.

@SimonSapin The difference between the two is that OPTIONS http://www.example.org:8001 asks for all available options of a server in general. The server may reply with Allow: GET, HEAD, POST, PUT. But if one requests OPTIONS http://www.example.org:8001/ it may send Allow: GET, HEAD since the user can't change the index page.

The difference is also explained in § 5.3.4 of RFC7230.

from rust-url.

I found that this can be easily solved without support by rust-url .

from rust-url.

I'm curious, what did you do instead?

from rust-url.