A Google programmer has made tools for Apple's Swift programming language available to developers using Microsoft's Windows operating system, a move likely to rekindle hopes that Swift, open source since 2015, will become popular beyond the macOS and iOS ecosystems.
On Tuesday, Saleem Abdulrasool, a software engineer at Google Brain who joined the Swift Core Team in January, announced the availability of a new set of downloadable Swift toolchain images for Windows.
Abdulrasool has been spearheading the effort to bring Swift to Windows for more than a year and now feels that the various components involved – the compiler, the standard library, and core libraries like dispatch, Foundation, and XCTest – have reached a point where early adopters can try them out.
"With these core libraries and the flexible interoperability of Swift with C, it is possible to develop applications on Windows purely in Swift while taking advantage of the existing corpus of libraries on the Windows platforms," said Abdulrasool in a post to the Swift blog.
The Swift team tipped its hand about Windows development back in January in a note from Apple's head of Swift development Ted Kremenek, and again in March when project developers discussed goals for Swift 5.3, which was released last week.
The debut of a Windows-focused forum on the Swift discussion website argues that Swift on Windows is now a thing, alongside Swift on Linux.
Official tailored Swift for Windows support promised in 5.3READ MORE
Abdulrasool has posted demo code for a Windows calculator app written in Swift as an example of the possibilities. And Readdle, maker of Swift-based cross-platform email app Spark, earlier this month, published a post about its efforts to develop a Windows version of its email client using Swift.
Enthusiasm for Swift outside of the Apple ecosystem has surfaced periodically in recent years, only to die down again. In December 2019, IBM, a booster of server-side Swift since 2016, backed away from the language. In January 2020, server-side Swift-hosting biz Vapor Cloud said it would discontinue its service in February.
Setbacks aside, Google sees reason to support Swift for TensorFlow code, as its TensorFlow developers explained in a 2018 document. Swift, it turns out, has some features and characteristics that prove to be particularly useful for machine learning, such as automatic differentiation. It's also significantly faster than Python when properly tuned.
Swift could see its reputation rise further still with the release of Swift 6.0, which doesn't yet have a specific timeline but should debut once the core team implements improved concurrency support and a resource ownership model similar to Rust's ownership system.
If Swift can match Rust's code safety potential, the language's more approachable syntax could help it win developers away from trendier Rust, and other high-performance languages like C and C++. That may also require high profile projects unaffiliated with Apple – the iPhone maker's imperious behavior and prickly relationship with third-party developers fosters a fair amount of suspicion that still colors perceptions about Swift. ®