Question:

This is a follow up question to: <a href="https://stackoverflow.com/questions/24291288/how-to-fix-cannot-infer-an-appropriate-lifetime-for-automatic-coercion" rel="nofollow">How to fix: cannot infer an appropriate lifetime for automatic coercion</a>.

I wonder why do these both structs differ in the way they are affected by lifetimes.

<strong>Example 1</strong>

use http; pub struct Request<'a> { pub origin: &'a http::server::Request, }

<strong>Example 2</strong>

use http; pub struct Response<'a, 'b> { pub origin: &'a mut http::server::ResponseWriter<'b>, }

They look pretty much similar to me except that the second one holds a mutable reference whereas the first one holds an immutable reference.

However, for <em>Example 2</em> I can't just use a lifetime for the reference. I <strong>must</strong> give a lifetime for the struct as well.

So, I wonder is there something <em>inside</em> the struct that causes such behavior or is it really because the one in the second example is a mutable reference. And if so, why exactly does that cause that.

&'a T means that you have a reference to a T object which is valid for the lifetime'a.

T<'b> means a T object containing an object inside itself valid for the lifetime 'b, as in struct T<'b> { t: &'b U }.

&'a T<'b>is thus a reference with lifetime 'a to a T<'b> object.

In the case of the ResponseWriter, it contains references to the Request and to the TcpStream, whereas the Request does not contain any references.