Avoid O(n^2) performance by reconsidering the full set of obligations only when we are about to report an error (#18208). I found it is still important to consider the full set in order to make tests like `let x: Vec<_> = obligations.iter().collect()` work.
I think we lack the infrastructure to write a regression test for this, but when I did manual testing I found a massive reduction in type-checking time for extreme examples like those found in #18208 vs stage0.
f? @dotdash
This adds impls of Eq/Ord/PartialEq/PartialOrd/Show/Default to Arc<T>, and it
also removes the `Send + Sync` bound on the `Clone` impl of Arc to make it more
deriving-friendly. The `Send + Sync` requirement is still enforce on
construction, of course!
This adds impls of Eq/Ord/PartialEq/PartialOrd/Show/Default to Arc<T>, and it
also removes the `Send + Sync` bound on the `Clone` impl of Arc to make it more
deriving-friendly. The `Send + Sync` requirement is still enforce on
construction, of course!
This allows unboxed closures that reference free type/region parameters to be monomorphized correctly in trans.
It was necessary to make `ty_unboxed_closure` carry around a `Substs` to accomplish this. Plumbing this through typeck revealed several areas where type/region parameters in unboxed closure types are possibly not being handled correctly. Since my goal was just to fix trans, I decided to leave FIXME comments on areas that still need attention and seek feedback on the best way to clean them up, possibly as a follow-up PR.
Closes#16791
This adds a `Substs` field to `ty_unboxed_closure` and plumbs basic
handling of it throughout the compiler. trans now correctly
monomorphizes captured free variables and llvm function defs. This
fixes uses of unboxed closures which reference a free type or region
parameter from their environment in either their signature or free
variables. Closes#16791
This PR changes the signature of several methods from `foo(self, ...)` to
`foo(&self, ...)`/`foo(&mut self, ...)`, but there is no breakage of the usage
of these methods due to the autoref nature of `method.call()`s. This PR also
removes the lifetime parameter from some traits (`Trait<'a>` -> `Trait`). These
changes break any use of the extension traits for generic programming, but
those traits are not meant to be used for generic programming in the first
place. In the whole rust distribution there was only one misuse of a extension
trait as a bound, which got corrected (the bound was unnecessary and got
removed) as part of this PR.
[breaking-change]
The arenas write the value to memory and then return a non-aliasing
reference to it. The returned reference can be mutable and can be
coerced to an immutable one.
[breaking-change]
