The most exciting thing that happened this week was that Eduard-Mihai Burtescu, more commonly known as eddyb, joined the language design team. Eddy has been involved in hacking on the compiler for years, and he has shown quite an astute sense for language design questions over the years. We are very excited to have him on board.
In terms of pending RFCs, we decided not to decide on the question of
#[repr(align)] just yet, since the proper interaction between
packed is still unclear. We proposed just
making it an error to combine them for the time being and settling it
We introduced the following RFCs into FCP:
Both of them are (directly or indirectly) targeting the topic of atomic operations applied across more types.
The following RFCs remain in FCP, mainly because of lack of time to discuss and consider them this week:
RFC 1492 proposes to permit
..notation in patterns as a more general replacement to
_, so that one can write a pattern like
Foo(x, ..)instead of
Foo(x, _, _, _).
RFC 1592 is a small amendment that corrects an oversight in the
original well-formedness RFC, by making tuples require that all
elements except for the last element be
Sized(just as structs do).
Much of the meeting this week was devoted to discussing and planning for next week’s development sprint – the compiler team and active rustc contributors will be gathering in mountain view for a focused week of hacking and planning. We are aiming to push the ball forward on both MIR adoption and incremental compilation, as well as plotting out how to revamp the compiler to be more performant and flexible.
A relatively quiet few weeks with regards to tooling and infrastructure, but some of the highlights were:
- A new approach has landed in the compiler and feedback would be most welcome!
target_envvalue for many Linux ARM targets has been normalized to just “gnu”.
- Nightlies for i686 FreeBSD Cargo are now available.
- A new
cargo installis now available to overwrite existing binaries.
- The RFC for Cargo workspaces has been accepted.
This week we discussed a large number of outstanding issues and PRs. We also reached decisions on the following RFCs:
FCP PR #1480:
RFC proposal for thread affinity
- Closed in favor of building this support out of tree.
FCP PR #1521:
Copy clone semantics
FCP PR #1542:
Add TryFrom and TryInto traits
FCP PR #1547:
as_millis function on std::time::Duration
- Closed in favor of out of tree development.
No libs RFCs are currently in FCP.