Worth noting maybe that although iOS 15 came out only in 2021, support for iOS 15 goes back to the iPhone 6S from 2015. Not very many people actively using iPhones older than that today.
The original poster asked about _fine-grained permissions_. Not about _runtime permissions_. Details matter.
Android did have very fine-grained permissions since first betas. Yes, they were install-time - a policy was generated at install time by the system, and the app itself was unable to change anything about it. Technically, it was a nice system, but users didn't understand that, they were asking for simplified model from iOS, so they got it in Android's 6.0 _runtime permissions_.
In the end, neither of these system (or: original Android did have it, but the simplified 6.0+ doesn't) has the most important permission: can an app talk to the network?
> Android did have very fine-grained permissions since first betas. Yes, they were install-time
They weren’t really that fine-grained from a user perspective. You could not accept/refuse individual permissions, you either accepted everything or simply not install the app.
iOS always had fine-grained permission in that you could grant/refuse individual permissions. For example: you could allow an app to access the camera but refuse location services. Android only recently gained that capability.
Even more important, iOS always put the permission request in context. If I install an app and it asks for a ton of permissions I have no idea why it needs them and if it makes sense for that app to have them. Why would a chat client need access to my photos ? But on iOS, I get that request the first time I choose to send a photo to someone. I immediately see by the context why it needs that permission and I can make an informed decision.
> They weren’t really that fine-grained from a user perspective. You could not accept/refuse individual permissions, you either accepted everything or simply not install the app.
They were fine grained: apps either had them in their manifest, or not. If not, they could not call the respective APIs without getting an exception.
Because there were so many, it would be a great burden to app developers to check for random mix of required permissions, whether it was granted or not. The complexity would shoot over the roof. When Android switched to runtime permissions, all the detailed permissions were grouped into fewer coarse ones; exactly because so much detail would be unbearable for both users (fatigue from the alerts) and developers (handling the enabled/disabled matrix).
As far as I remember, iOS originally didn't have any permissions. It got them once certain app was stealing users address books, so it got confirmation for accessing contacts, camera/photos and a third thing that escapes me at the moment (location?).
> But on iOS, I get that request the first time I choose to send a photo to someone. I immediately see by the context why it needs that permission and I can make an informed decision.
The app can also remember that it got the permissions and do the nefarious thing behind your back. While it may look better, it is really not; it also won't work if the permission system is fine-grained: too many types of permissions and users will get lost. Also, users accidentally pick the wrong choice and then wonder, why the app doesn't work like they expect, or how to change it.
