Friday, May 30, 2025

Xcode 16.4

Apple (downloads):

Fixed: Users may see excessive CPU utilization from diskimagesiod which reduces simulator performance, increasing boot time, process launch times, and test execution times.

[…]

The command devicectl diagnose now obtains a sysdiagnose from your Mac and all available devices by default.

[…]

Fixed: NSURLSession was always timing out and failing in iOS 18.3 simulator runtimes.

Fixed: Some C++ headers were experiencing crashes in syntax highlighting and Quick Help.

Not a lot of changes here. It seems to be working the same as Xcode 16.3 for me—no new problems, but it doesn’t seem to fix all the issues introduced in 16.x, either.

Sarah Reichelt:

I see some people reporting problems with Xcode 16.4 but it solved a problem for me. I was holding off the pre-release of my book “Escape from Tutorial Hell” because 16.3 had a bug that crashed a playground the used JSON decoding. So please don’t try to use Xcode 16.3 with any playground in the book. 16.2 and 16.4 are both fine.

Previously:

6 Comments RSS · Twitter · Mastodon


Let's say the rumor is right and all of Apple's OS releases go to 26. Or at least x,0 this fall.... how does anyone feel about Xcode going at the same release level of the OS releases?


They shipped another broken Simulator, this time due to the use of WKWebView: https://developer.apple.com/forums/thread/785964


@Dave sounds like they've got way bigger problems that what number they choose to label it with.

The way I feel about it is that if they try to change the version number it calls itself it may just break completely.


@Bart, thanks for answering my snakiness. I mean that sincerely. Here's my more constructive take....

-- @Arek, I can't speak of a broken Simulator. My area is CoreImage, which 2 versions ago was supposedly fixed to *finally* work with a simulator but I've never been able to get working. So I use physical devices. Funny thing though, in order to get the latest API, I am forced to download the Simulator for it. Could be wrong... but Xcode updates assume I need it and never found a way to bypass the S-L-O-W "optional download after the almost equally slow "installing portion of the Xcode update.

-- Speaking of that, About three years ago this annual cycle of major betas they forked iOS and iPadOS. Maybe there's a big picture plan in place, but I have yet to see it. Apparently so it is with Apple support, as I was told (by an "engineer"?) back then that while they rolled out iOS and Xcode - and waited a few weeks for iPadOS - this, after breaking my ability to put all my iPads into beta mode.

-- Odd, this last annual cycle the updates have been almost every 8 weeks for dot (not to be confused with dot-dot) updates in OS, and Xcode not so frequently - yet it feels almost like that. Not because of Swift (that's a totally different topic) but... my guess is AI (not "artificial" but "Apple" intelligence). It *has* infected Xcode - the default option, at least for 16.4 (there is go @Bart, in order to use my intelligence I had to go to the About screen to make sure) is for "predictive" coding... that's for lazy coders who can't do it on their own - to be on. At least I can untick that post-download checkbox. (Never figured out why I can't the macOS one.)

-- Warnings. Here's one "CLIENT OF UIKIT REQUIRES UPDATE". It goes on to say that that it will become an assert at some point in the future, breaking my apps because I haven't fully adopted the UIScene lifecycle yet.

-- Finally, - and this is for @Bart, whom I agree with his comment - consider this code:

if #available(iOS 11, *)

Make it iOS 13. Of maybe iOS 15 - I think that's when Apple deprecated OpenGL (but at least gave gave me a way to squelch that warning through Build Settings). So let's go with iOS 17. Why if in my app settings when I choose iOS 17 does it automatically go to 17.6? I never figured out how to just say 17.0 and just accept it. But back to the simple line of code, checking what version the *device* is running on. Absolutely, if Apple changes their version numbering it will be a mess. But it will be - to quote a unnamed politician - short term pain for what could be long term gain.


@Dave any time, thanks for being a good sport.

I've got no answers for you, I just know version numbers have been a whole interesting side problem for a long time, and nobody except marketing likes shaking them up.

I know this was a problem on Windows and Microsoft just quietly kept incrementing the "real" version numbers for programmers to check that. Good thing too when they come up with names like "Windows 10 2003" not to be confused with "WIndiows Server 2003".

I was only half joking. Apple seems to be in such a mess with their developer relations and their internal system development, that I hope they will announce a simple version number check because they've already made everything else overly complicated.


If I were a betting man I'd go with this... names! You want them, you get them! One thing - downloads of macOS will be named by country landmarks where possible (if Trump can tariff an island that only has a penguin population, well) and for the US downloads? We'll go one name per state. I simply cannot wait for macOS Presque Isle 17.0. (End of sarcasm.)

Leave a Comment