MMS • Sergio De Simone
Article originally posted on InfoQ. Visit InfoQ
The Swift language workgroup has detailed the main focus areas for the development of Swift in 2023 and further on, which include defining an ownership model for memory management, non-copyable types, a macro system, and C++ interoperability.
Ownership is an approach to memory management that has recently become popular thanks to Rust, where it holds as one of the most defining language features and at the foundations of its ability to provide memory safety guarantees.
Rust uses a third approach: memory is managed through a system of ownership with a set of rules that the compiler checks. If any of the rules are violated, the program won’t compile. None of the features of ownership will slow down your program while it’s running.
Discussions on the introduction of ownership in Swift began back in 2017 and led to the definitions of an Ownership Manifesto.
While work on ownership did not lead anywhere, the Swift language workgroup is bringing it back with the aim to provide programmers with more control on values in memory. This could include prohibiting implicit copies, enabling ownership transfers, and borrowing values without copying them. Additionally, Swift could get support for non-copyable types to restrict the lifecycle of critical values.
These controls will enable new ways to work with data in memory, combining the performance of current “unsafe” constructs with the safety of Swift’s standard library features.
Another promising area for Swift is the creation of a procedural macro system to enable the creation of advanced libraries and DSLs. Macros are a mechanism for code generation that applies a transformation to source code at the lexical, syntactical, or semantic level. A number of existing features in Swift that could be implemented as macros are the Codable
protocol, string interpolation, property wrappers, and result builders. According to the Swift language team, using macros to build new features like those listed above would free up more resources for alternative work on the language and its tooling.
While work on Swift macros is still in very early stages, there are already some ideas about what they could look like in Swift. Here is an example of a possible stringify
macro:
macro(contexts: [.expression, .parameter], external: "MyMacros.Stringify")
func stringify(_ value: T) -> (T, String)
C++ interoperability aims to enable calling C++ code from Swift as well as calling into Swift from C++ code. It is the feature that will probably make it first to an official Swift release given its more advanced status. The current implementation of C++ interop already supports owned value types, trivial value types, foreign reference types and iterators, and provide an answer to fundamental questions around methods, pointers, and l-value and r-value references, according to the language team.
While the three areas delineated above are the most innovative, Swift evolution will also focus on bringing forth existing features, such as concurrency and generics.
In particular, for concurrency the goal will be improving data isolation provided by Sendable and actors, whereas work on generics will bring support for variadic generics, i.e., generics that have a variadic number of placeholder types.
No schedule has been provided for the new features described here and it is not clear whether they will be available in Swift 6 or later. InfoQ will continue to report about Swift development as new information becomes available.