Roadmap 2.0

[nia-e]

See #48 for the previous (now-outdated) roadmap.

Core traits, types, and their interactions

• Merge proposed StaticAllocator trait & changes to Allocator
  • allocator: refactor for stabilisation rust#157428
• Determine the soundness situation w.r.t. Clone interactions
  • Proposal: unsafe trait AllocatorClone: Allocator + Clone
  • Proposal: conditionally-unsafe trait impls
• Merge a NativeAllocator<A: StaticAllocator> lang item for LLVM intrinsic interop & noalias emitting
• Explore other proposed extension traits
  • Proposals came up for PartialEq marking allocator equivalence - do we want something like this?
  • Deallocator?

Beyond GlobalAlloc

• Merge a nightly GlobalAllocator trait
  • allow Allocators to be used as #[global_allocator]s rust#157153
• Deprecate/hide GlobalAlloc over an edition
• Implement Global in terms of NativeAllocator & stop having it be a lang item

Expanded collection types

• Allocator-generic String
• Explore options for more fallible methods on collections
  • Proposal: Bound infallible methods on alloc types behind A: InfallibleAllocator libs-team#797
  • Proposal: Tracking Issue for alloc::collections::fallible::Vec rust#157392
• Move collection types into core

Allocator-aware pointer types

• Make Box no longer fundamental over the allocator
  • Make #[fundamental] only apply to the first argument of Box rust#160162
• Stabilise the allocator on Box
  • Doing so is a stable commitment; see also further context
• Figure out semantics and trait bounds for Arc/Rc<_, A>
  • Proposal: Store the allocator inline with the value instead of in the Rc/Arc itself

Stabilisation path

• Stabilise the core Allocator trait & minimal functionality around it
  • alloc: stabilise Allocator rust#156882
• Plan out a roadmap for further stabilisations

Meta

• Land an allocator project goal
  • Propose allocator project goal goals#752
• Clarify and expand documentation on the allocator trait family
• Clean up old & now-irrelevant issues in this repository