join_all() is too picky about lifetimes

[dwrensha]

Consider the following function:

fn foo<'a>(bufs: Vec<&'a [u8]>) -> Box<::futures::Future<Item=Vec<usize>, Error=()> + 'static> {
    let iter = bufs.into_iter().map(|b| ::futures::future::ok::<usize, ()>(b.len()));
    Box::new(::futures::future::join_all(iter))
}

I expect this to typecheck successfully, but I get an error:

error[E0495]: cannot infer an appropriate lifetime due to conflicting requirements
  --> src/main.rs:35:21
   |
35 |     let iter = bufs.into_iter().map(|b| ::futures::future::ok::<usize, ()>(b.len()));
   |                     ^^^^^^^^^
   |
   = note: first, the lifetime cannot outlive the lifetime 'a as defined on unknown free region bounded by scope CodeExtent(212/CallSiteScope { fn_id: NodeId(122), body_id: NodeId(276) })...
note: ...so that types are compatible (expected std::vec::Vec<&[u8]>, found std::vec::Vec<&'a [u8]>)
  --> src/main.rs:35:21
   |
35 |     let iter = bufs.into_iter().map(|b| ::futures::future::ok::<usize, ()>(b.len()));
   |                     ^^^^^^^^^
   = note: but, the lifetime must be valid for the static lifetime...
note: ...so that the type `futures::JoinAll<std::iter::Map<std::vec::IntoIter<&[u8]>, [closure@src/main.rs:35:37: 35:84]>>` will meet its required lifetime bounds
  --> src/main.rs:36:5
   |
36 |     Box::new(::futures::future::join_all(iter))
   |     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

error: aborting due to previous error

The problem seems to be that JoinAll<I> is parameterized by the type of the IntoIterator that is passed into join_all(). In the above case, that type is not 'static, even though its Item is 'static.

[carllerche]

Is there an action item to take from this? The issue has been inactive for a year.

[stuhood]

@carllerche : I believe this will be useful for 0.2, but requires changing the parameters of JoinAll see the linked PR and the other issue.

[alexcrichton]

@carllerche the fix, a breaking change, is at #286 (linked above)

[stuhood]

Both this and #286 should be closable now, as they've landed in the 0.2 branch.

[alexcrichton]

Thanks @stuhood!