why the minimum is iOS 11?

Is there any particular reason for this? I can’t find anywhere that the minimum target deployment for apple is now iOS 11, and there are ton of people on iOS 10 still.

hello? anyone? 2019.3 moved from ios 9 to ios 10. Ok, I can understand it. But 3 months later and it moves from ios 10 to ios 11 when there are a trillion devices there stuck with ios 10 and not able to upgrade?

Please, explain why ios 11 must be mandatory to compile on unity 2020 and ios 10 is not enough. It will leave way too many people out of Unity games using the latest version.

What are those devices?

AFAIK iPhone 5 and 5c are dropped and iPad 4th gen, and… that’s about it?

Meanwhile this allows them to drop GLES 2.0 completely, since all iOS 11 devices support Metal.

2 Likes

Also, in addition to what @AcidArrow said, 2020.1 is still in alpha, so not meant for production yet, and should only be actual released (out of both alpha and beta) next year between end of Q1 and start of Q2 (if Unity follows the same release schedule since 2017).

2 Likes

There are tons of people playing games on ipad 4th gen. It’s a very common device.

1 Like

Not really? https://david-smith.org/iosversionstats/

But I’d welcome stats that show otherwise.

Also, that specific stats page, since it is from an audiobook app, they are much more weighted towards the low end than the average Unity game (at least from back when Unity had their stats public so I could compare)

1 Like

yes really. I’ve seen those stats too but here is the thing… that’s not app depedent.
I mean, if you wrote an app only for ipad and not for iphone, you would say: hey, 95% of iphone users are upgraded. True, but that’s not your target.

Here is a similar situation. You can’t just look at ipad and say that ipad 4th is only 1% of the ipad loading your app, therefore, almost no ipad 4th. It could well be that YOUR app is mostly used by ipad 4th. Or that nobody uses your English reading app on older ipads in United States but worldwide it’s flooded with that version because it was a cheaper alternative. What I’m saying is that many, many, many ipad users are stuck on ios 10 or lower. Perfectly fine working ipads specially outside the United Sates. If you enforce ios 11, you are going to leave ton of people out of any upgrade. I don’t understand why 2019.3 enforces ios 10 and in less than 3 months, you are going to enforce ios 11. I would give it some more time. Not that my opinion matters, but hey, I always try to make it work on as many devices as possible unless they are crap. And ipad 4th works perfectly fine for almost any game nowadays. It’s not like an iphone 2g.

I understand, but if you care that much about supporting the iPad 4, stay on 2019.4 LTS for a while. You still have more than 2 years of updates for that and it’s perfectly doable.

I can’t talk in place of Unity obviously, but I’m thinking dropping ES 2.0 support simplifies things for them and allows for more features to be universally supported on iOS.

3 Likes

In terms of raw power, its not in the same state that the early iphones quickly found themselves in.

But the ipad 4th gen is a 32 bit device, thats why it faces a ‘premature death’ - lots of modern apps wont run on it as a result, apple have moved devs etc onto 64 bit only. I have a friend who hates to upgrade but even he had to move on to a 64 bit ipad this year, he just could not stand to be unable to run lots of new apps on it.

Yeah, this had me intrigued, so I looked at all stats I could find and everything seems to indicate that less than 5% of devices use <= iOS 10. Likewise, any stats showing otherwise very welcome, the more data we have access to as developers the better.

ES2.0 is the worst thing in existence for mobile graphics developers. It’s basically the annoying person who won’t leave the store even though the lights have been dimmed a while.

Then says “was just browsing”.

2 Likes

lol, yeah I also wish ES2.0 was dead and 3.1 was the standard for mobile. Unfortunately on Android is still a huge market. I have so much shader work that I would love to use but I can’t unless I remove 1/3 of the userbase.

For iOS there’s a thing about PVRTC. Compression artifact looks horrible around alpha edges and it requires square POT dimension as opposed to just POT, but default “Compressed” in Unity falls back to that thanks to having to keep backward compatibility.

Trying to make the game works as many devices as possible sounds ideal, but it costs someone with better device a bad experience in exchange. I view that as not “works”. You could as well view that the game is now somewhat incompatible with newer devices even though number of devices that could download the game goes up, number of devices with optimal experience went down.

If we guarantee Metal, then the default could now be ASTC instead or even ETC2 then we can crunch it. Artifacts look subtle and we can optimize with block size, with quality (how channels are summed up) and get non-square POT dimension. We can finally stop worrying about someone coming without Metal (< A8). I am very glad that < A8 devices is finally being phased out. I just cut down iOS download size from 50MB to 30MB from this decision to ignores PVRTC alone.

Researches :

3 Likes

Was ES 2.0 dropped also for Android in 2020.1?
I checked few months ago and seem to me there are still lots of ES 2.0 devices out there and many with poor ES 3.0 implementation.

I don’t know, but I don’t think so.

On May 2019 GL2.0 devices take up 21.1%
https://developer.android.com/about/dashboards/index.html#OpenGL

1 Like

And that is little? To me it’s a lot of devices.

It’s not little, but we are talking about iOS.

1 Like

I do think it is a lot numerically but there are more considerations to weight than just applying percentage to # of human if your target is to maximize revenue :

  • You want a maximum number of devices able to download the game no matter what, so you predict that it translates to maximum revenue. Then go for this.
  • Your own development velocity : if you can ditch them and make a better code and put out faster new features? Or eliminates more bugs? May turns out to be better profit than maximize compatibility.
  • How many of 21.1% are actually your potential customers or better, paying customer : better confirmed with your own game’s analytics that could report GL version. (Or better generic data from countries that more likely game, like Japan/Korea) Maybe they came from devices which could not handle games in the first place, or in a hand of person not interested in gaming. You can deploy multi APK with GL3 priority over GL2, and see how many got that GL2 APK fallback. (I have not done this, already shifted to AAB with only crunched ETC2 for dev simplicity. Same for iOS with ASTC.)

In June 2014 GL2.0 was 83.6% vs GL3.0 16.3% (https://stackoverflow.com/a/13036753/862147). I would not think like this in that time. My threshold would be around < 30%, maybe yours is lower like < 5%. Wait a few more years in that case.

2 Likes

The problem with opengl 3.0 is that i’ve heard lot of horror stories on android. Things going half the speed than opengl 2.0. So it seems that the only “safe” way for android is opengl 3.1 which has still a very little user base. So no way to ditch 2.0 so far.

On a side note. Is there an option to change the default image compression type to ASTC for ios? Since we can finally remove the horrible PVRTC but manually going through every image on a project would kill me