What is the point of link rel=“self” in a REST API?

Question

I often see the following in HTML documents

<link rel="self" href="http://example.com/something">

or like this in JSON

 link: {
     rel="self",
     href="http://example.com/something"
 }

or in XML

 <atom:link rel="self" href="http://example.com/something" />

So I had some questions:

1. Why include this link? What advantage does it bring? (Please tell me there is a reason to it and its not just a "good practice" talisman)
2. How should I exploit this link in my clients? What are the use case for this link?
3. When shouldn't I use this link? When is it pointless to include it?

Answer 1

It is a self reference, so the client will know that the IRI (http://example.com/something) is an identifier of the resource the representation is about.

It can be important when your resource can have multiple identifiers, for example http://example.com/users/1 and http://example.com/users/1?fields="name" can identify the same resource, but a GET on them can result different representations.

By media types like HAL you use this to identify embedded resources as well. For example:

{
    "nick": "John",
    "_embedded": {
        "cars": {
            "items": [
                //...
            ],
            "_links": {
                "self": {
                    "href": "http://example.com/users/john/cars"
                }
            }
        }
    },
    "_links": {
        "self": {
            "href": "http://example.com/users/john"
        }
    }
}