> it's possible to ship fixes for a new iOS version before that iOS version actually launches

No it isn't. Apple doesn't permit code linking against the latest SDK to be submitted until the SDK goes "GM". That happened today.

To expand on this, if your app is "broken" because of changes made that require new OS features in order to "fix" them (such as the new limited permissions for photos, contacts, and location), you were completely unable to issue a new update to fix them until the end of the event. That gives you ≈24 hours to get the new GM, create a build (including setting up your CI system with the new GM, if applicable), make sure no new bugs have been introduced, submit to and get approved by App Review, and launch the update. If you couldn't get all that done in that time, you're probably going to have at least a few negative app reviews from early adopters.

And even then, who is to say that users are going to instantaneously download your update before/during their upgrade to iOS 14 (even if they have automatic app upgrades turned on, which they might not).

Being able to point your users to an update to fix their problems when they come to you for support, or when they go to write an angry review that their app is crashing but they see “iOS 14 support” at the top of the app’s page, is likely to help there. Especially since the earliest adopters are frequently technically competent enough to do that but also frequently demanding enough that their apps work on day one.

So, maybe you can get to it in a couple of days. Fix what you can using the old SDK before GM build becomes available. Then you have a small amount of updates once apps using the new SDK can be submitted. The app will be temporarily broken for a few days on a brand new OS until the update goes live. Is that really a big deal? The world doesn't come to an end if you get a few bad reviews from impatient users. I remember being in the mobile app development rat race and am so glad I left that world behind. Too much daily panic about things that really don't matter much.

Ok, but surely you can understand why people would be upset with Apple for doing this when they had no reason to do so?

The new location precision user auth seems to have been done in a backwards compatible way, at least in my testing:

- app with existing location permission in iOS 13 gets precise location.

- when the user prompt appears, it looks like precise location is the default in the UI.

(This is with an app build against iOS 13 SDK -- not sure yet if the iOS 14 SDK changes things.)

I think Apple will likely change this in the future, but right now devs are getting the time they need to adjust.

(As is, this new permission does almost nothing since 99% users will accept the defaults. I think they'll probably require more explicit action on the user's part to get precise location in the future.)

That isn't strictly true, several apps have launched their iOS14 ready versions ahead of the public release. All they can't do is use new SDK features.

You can't link against the new SDK, but you can link against an old SDK, but fix the bugs which are associated with iOS 14.

~All of the bugs I've ever seen on a new iOS release are just where they've tightened up some behaviour.

You say "tightened up" and perhaps that's what they meant internally, but as it stands my (simple) app's behavior when using the search bar is different and broken on the new update even though I'm using the API in a supported way. So it's really a bug.

I have applications and games from the 90s that still work in Windows 10. The idea that developers have have to constantly push updates is unproductive in my opinion.

A ton of software from that era stopped working when XP was released. For instance storing settings in an application's installation folder was common until Microsoft stopped it. Anything that ran as a service but interacted with the desktop was out of luck. This idea that everything from that period continues to run is a myth.

I never said everything. But a lot of the software still works. Some work with minor workarounds. Of course some programs will break with an update. But in my opinion creating a culture where an update is expected for software that is a couple of months old at best is really unhealthy.

no? all these things are just a couple configuration away from working.

the real things that were dropped since then were 32 bit protected mode apps, and for such you have plenty options for both virtualizing and paravirtualizing.

I can boot basically any other version of Windows under my windows system using Microsoft provided software.

I can even just go boot up older version of Windows on current hardware if needed be, up to a point.

almost everything from eighties to today I am able to run one way another, nothing "stopped" working.

Apple physically removes old software from their delivery platform or signing servers and if you lose your devices it's lost forever

how does it even compare?

That's a decision made by Microsoft which has pros and cons. There's significant overhead in maintaining this backwards compatibility. For an OS used by corporations all over the world, its worth it. For a mobile OS it will likely reduce their ability to deliver changes.

Apple used to give a shit there. They did a bunch of extra work so you could run OS9 apps under OSX for a few years, and they did an amazing job of making 68000 code run on PowerPC...

Wait until you see Rosetta.

I agree, I think a lot of what holds Windows back is the baggage from the past. Ideally OSes would live somewhere in a healthy middle, where we can’t expect 10 year old software to run on the current version of an OS. On the other hand we should expect 1 month old software to run on after an OS update.

Until you read Raymond Chen’s blog about all of the hacks that have been built up for the sake of backwards compatibility or try to run Windows on a low resource computer like the low end Surface computers and can’t because of code bloat.

The new OS version is a moving target, and it only really settles down with the final release candidate - which can still be changing a couple of weeks before the actual release.

So yes, it's possible to write to the beta, but no, it's not possible to guarantee there will be no last minute breaking changes.

I have been developing an app using the new SwiftUI 2.0 for iOS 14 during this beta. My builds have been breaking constantly between beta seeds, due to symbol name changes etc. And Apple have never allowed you to upload a binary linked with a pre GM SDK for release.

But they only released the XCode GM today.