By focussing on GLES2 - are you not losing too many players?

[trjp]

Having done a bit of digging, I think the figures Nicolas is touting for Android 2.2+ uptake are a bit optimistic.

A lot depends on the sort of players your games attract, of course. I just went through the stuff I've done (that I have a decent amount of stats for) and the % of players who are on 2.1 or earlier isn't 11% or anything like that.

It varies between 22% (double Nicolas' number) and a little over 30%!! That's a BIG chunk of players I'd be cutting-off by moving to GLES2 (or developing from scratch using it) - is that entirely wise?? (I think not)

I realise those on older versions can keep playing the older versions of existing games, but by moving support and new development to GLES2, you're cutting-off a lot of customers.

Yes, most people have an update route to 2.2 at least - but clearly a lot of people aren't bothering. I don't really have enough data to see if the 2.2+ percentage is growing at a significant rate but I imagine it's certainly growing.

A big issue with the numbers are, of course, tablets - those which run 3.0 are obviously distorting the data upwards. I tend to ignore tablets - mainly as I think most current Android tablets are overpriced junk - and that perhaps explains why I have a higher % of 2.1 or earlier customers.

I thought it was worth pointing-out anyway - it's a busy marketplace already - you'd be foolish to turn-away too many potential customers??

[ltmon]

My raw stats over 3 apps are similar (25%-ish).

However, it may be that it's including a lot of devices that people have retired and replaced, but not unregistered or uninstalled the app from, hence the discrepancy?

The market stats are a bit opaque in what they are actually measuring, so I would probably trust Google's advertised 11% more than than the stats I get. It's probably closer to actual current sales trends rather than the more historical stuff we get in the console.

[trjp]

I used stats from several games (one of which is only a few weeks old so won't have too many 'older' users) - and I used stats from the Android Market and other sources so I think it's probably a fair reading of the % of 'my' likely users - it may not resemble other people's of course.

In 12 months time things will be very different I'm sure (hopefully the tablet marketplace won't be a mess if nothing else!!) but right now I think you realistically have to think about GLES2 locking out more like 20%+ of your potential customers rather than 11.

[bond]

We can't judge the real numbers out there by the stats our apps have...
The number will not be the same because our pool of devices is regarding people who downloaded our apps not the overall devices available.
I will tend to agree with google's overall numbers more than our own stats on this matter...

Android 1.5 Cupcake 3 0.6%
Android 1.6 Donut 4 1.1%
Android 2.1 Eclair 7 8.5%
Android 2.2 Froyo 8 30.4%
Android 2.3 -
Android 2.3.2 Gingerbread 9 0.6%
Android 2.3.3 -
Android 2.3.7 10 54.9%
Android 3.0 Honeycomb 11 0.1%
Android 3.1 12 1.5%
Android 3.2 13 1.7%
Android 4.0 -
Android 4.0.2 Ice Cream Sandwich 14 0.3%
Android 4.0.3 15 0.3%

[trjp]

If you think Google's overall numbers are better than the actual numbers your (or my) apps gets that's upto you - but I'd have thought using your own data is always better?

Google has worldwide reach but many people don't translate their apps into every language and so see restricted sales on some countries (the obvious BIG examples being Korea/China etc. who most western developers don't bother considering/translating for and vice versa).

Devices available around the world vary too - in the West and Japan you'll see a lot of high-end devices but in other countries you'll see more low-end cheap/non-Google approved cheapo tablets and phones (which are likely running older versions of Android).

Conversely tho, people in the West are now getting Android handsets 'free' - they didn't even want a smartphone but the Galaxy Ace etc. are now the default handset given with basic contracts (such people won't bother upgrading or whatever - but are potential customers nevertheless).

Even the type of Apps you're making will have an effect. People with high-end handsets/tablets will often look for apps which will "make the best use" of them (e.g. they'll ignore the 'simple looking games') - wheras people who like word games or maths puzzles may not give a damn and be happy with older devices and won't necessarily have (or need) an ICS MultiCore device!!

I tend to think that unless GLES2 offers you features you can't get otherwise or performance benefits you NEED, you'll be foolish to exclude the devices which can't run it!?

[j_box]

Well,
GLES2 requires OpenGL 2.0 which is supported by is 90.5% of market.

Then we compile AndEngine-GLES2 under Android 2.2 (actually if you check the source files of the engine you'll see that it requires Android 4.0.2 which is supported by only 0.6% of devices, but I usually do manual removal of all references to the Android 2.2+++ SDK and compile AndEngine-GLES2 under Android 2.2.) which is supported by 89,2% of devices.

Assuming the worth case scenario: that all "bad" devices are extremely badly affecting our statistics by containing OpenGL 2.0 without Android 2.2 or containing Android 2.2 without OpenGL 2.0 (which is not true, because Android 2.2+++ devices tends to use more OpenGL 2.0). Then we arrive to ~79.7% ratio of devices which contains OpenGL 2.0 & Android 2.2

So, finally: OpenGL 2.0 & Android 2.2 is used by at least ~80% of Android Market users by the 29 January 2012.

[trjp]

20% is probably a fairer figure than 11% - and that's 1 in 5 of your potential customers which is quite a bit in anyone's book?

As I said earlier, if you need the new features/performance gains then it's probably fine but moving "just for the sake of it" seems unwise?

It's a shame, because using AndEngine as a library is much better than the old JAR setup - but that seems to be exclusive to the GLES2 release?

[RealMayo]

It's a shame, because using AndEngine as a library is much better than the old JAR setup - but that seems to be exclusive to the GLES2 release?

Actually Nicolas told me that GLES1 should be used as Library projects as well. When I made my GLES1 tutorial I didn't fully understand Library projects at that time, so I favored Linked projects instead. Nevertheless, you can use Mercurial in Eclipse to retrieve the GLES1 extensions and then implement them as Library projects.

[j_box]

20% is probably a fairer figure than 11% - and that's 1 in 5 of your potential customers which is quite a bit in anyone's book?

I don't really think this 20% are anyones potential customers. They are some weird users from the stone age where the fire was just invented.

[trjp]

20% is probably a fairer figure than 11% - and that's 1 in 5 of your potential customers which is quite a bit in anyone's book?

I don't really think this 20% are anyones potential customers. They are some weird users from the stone age where the fire was just invented.

You can, by all means, take that (unfortunate, IMO) attitude - but their money and (ad viewing) time is as-good as everyone else's - and one of the reasons Android is gaining market share is that allsorts of people are getting (often without even seeking them out) Android devices...

Seems silly not to offer them the chance of enjoying your work?