I updated my shell script for listing upcoming/experimental feature flags in the Swift compiler. The script now outputs CSV, which allows you to pipe it into other tools for filtering/sorting/pretty-printing. #SwiftLang https://gist.github.com/ole/478874632fca61869928a0cc0a956972
Level up your Swift skills:
9:30 AM - noon: “Level up your Swift” with @twostraws 1 PM - 4 PM: “Concurrency Fundamentals with Swift 6.2” with @mattiem 
Sunday entry only for workshop participants.
Lunch social at noon to connect & recharge!
RSVP: https://omt-conf.com
Apple Weather App (Benja’s Version)
A minimalist take inspired by the original Apple Weather app.
Check it out on GitHub: github.com/benjaminrsalgado/weather
Sometimes you gotta sherlock yourself. I think that Swift Testing should have something akin to Nimble's Polling Expectations.
https://forums.swift.org/t/pitch-polling-expectations/79866
Also, wow, macros are hard. I am so glad I don't have to write them on a day-to-day basis.
Not sure whether this is a #SwiftLang bug or not, but I've run into this. You would think T should be mutable always, because reference type.
#Swiftlang friends, what's your favorite complex Package.swift file?
Want to show one that assembles different targets per OS etc., and see whether we can learn from that in an OSS project.
Interesting that
guard let self = self
on a weak self isn’t recognized by the error reporting mechanism as a pointer to self being captured in a closure, only the actual self is.
![Swift source code:
class Foo {
let value = "Hello!"
func callAutoClosure(innerClosure: @autoclosure (() -> String)) {
print(innerClosure())
}
func weakSelfCall() {
let outerClosure = { [weak self] in
guard let self = self else { return }
callAutoClosure(innerClosure: "The value is \(value)")
}
outerClosure()
}
func strongSelfCall() {
let outerClosure = {
callAutoClosure(innerClosure: "The value is \(value)")
}
outerClosure()
}
}
And an error message on line 19 that calls callAutoClosure() with a strong self “Call to method 'callAutoClosure' in closure requires explicit use of 'self' to make capture semantics explicit”.](https://assets.chaos.social/media_attachments/files/114/461/687/746/849/595/original/ead6f1027f96a68d.png "Swift source code:
class Foo {
let value = \"Hello!\"
func callAutoClosure(innerClosure: @autoclosure (() -> String)) {
print(innerClosure())
}
func weakSelfCall() {
let outerClosure = { [weak self] in
guard let self = self else { return }
callAutoClosure(innerClosure: \"The value is \(value)\")
}
outerClosure()
}
func strongSelfCall() {
let outerClosure = {
callAutoClosure(innerClosure: \"The value is \(value)\")
}
outerClosure()
}
}
And an error message on line 19 that calls callAutoClosure() with a strong self “Call to method 'callAutoClosure' in closure requires explicit use of 'self' to make capture semantics explicit”.")
Trying to understand what is the best/proper way of injecting a ViewModel to a #SwiftUI View, they both work but which one is best and why? I’m using the Observable macro in the ViewModel
New blog post! 
How to add dependencies to a binary Swift package. I have looked all over for ways to make things happen and have finally made it work for pulling in an external binary package as a proper dependency.
https://danielsaidi.com/blog/2025/05/02/adding-dependencies-to-binary-swift-packages
I'm about 30 minutes late, but here's the first release of ATCommonTools.
I'm not going to put it into Swift Package Index until all of the relevant TypeScript objects have been converted, the DocC is completed, the README looks better, and the package in better shape.
#Swiftlang #ATProto
GitHub - ATProtoKit/ATCommonTo...
Just pushed the first draft of the landing page for my new habit tracking app! Built with Astro on the front and Vapor on the back for waitlist.
Check it out and let me know what you think 
https://habit-app.sakrist.com/
I don't get this:
People say "you shouldn't make your Foo protocol itself Sendable most likely, but instead require Foo&Sendable in dependent code" one minute, and the next propose "use an actor, you just need to make all of Foo's methods async for actor isolation support"
New Proxygen version is now available. Almost every part of the app has been updated. Perhaps the biggest thing is that there is now a direct download at https://freshbits.fi/apps/proxygen/.
Right now I’m leaning toward just wrapping the whole type in `#if os(iOS)`.
I have library targets that support both iOS and watchOS, with some types marked as unavailable on watchOS using the `@available` attribute. I'm curious why I still need to wrap iOS-specific APIs inside those types with `#if os(iOS)`, or the target won’t compile for watchOS.
SK8 has some clever ideas, like VirtualType (basically retroactive protocol conformance on classes you don’t own like in #SwiftLang), but it also does some things AppleScript luckily rolled back.
You create new handlers using a dialog panel, like you'd name a new file. That's slow. Also, a handler in SK8 looks like:
on mouseDown of me (a Board)
drag me with live
end mouseDown
So they threw out the English syntax right from the start (vs. AppleScript, where they just don't enforce it)
3/6
How do you deal with OSLog/Logger not logging in #Xcode Previews, only the print function?
I'm currently using this 
but it's boring. And the OSLogMessage interpolation makes it hard to wrap it.
Any better way to do that?
In Swift, is there an accepted "deinit()" practice (e.g., free up any allocated resources) for structs when they should be destroyed, or just create some "cleanup()" function that can be called? I'm thinking along the lines of releasing network resources or GUI elements.
Or is this a use-case for a class?
Almost missed it - today is @SwiftPackageIndex‘s 5th “first commit date”! To celebrate, we finished adding full support for #SwiftLang 6.1 and have started processing thousands of Swift 6.1 builds 
Thanks to @daveverwer for the great collaboration and to all our supporters and sponsors. It wouldn’t be possible without you! 
