This graph shows the challenge that faces Android developers; over 18,000 distinct devices can immediately use their app, making optimization a real challenge. Compared to last year, fragmentation has risen by around 60%, with 11,868 devices seen last year compared to 18,796 this year. Compared to our first report, in 2012, device fragmentation has more than quadrupled – with our first report showing that there were ‘only’ 3,997 distinct Android devices. Another way of looking at how fragmentation affects developers is to see what % of the market is occupied by the top 10 devices – as developers generally only own a few devices to test on. Last year having the 10 most popular devices in the market would represent 21% of the devices out there, this year that number has fallen to 15% – making testing on a few devices less representative.
Some very interesting numbers from OpenSignal.
One small problem with those numbers and percentages is that most of those devices just aren’t relevant, at least not to American/Western developers.
Phones more than two years old? Ignore them,
Phones from China and India that aren’t sold retail in the US? Ignore them. Only enthusiasts or total cheapskates will use them, and they’re not relevant.
(DealExtreme provides an interesting view here; there are 64 pages – of 30 phones each – of cheap no-name Chinese phones you can get from them.
[The middle of page 26 is where it starts to transition to
historical not-in-stock phones; this indicates something like 770
available who-cares Android phones.]
Not a single one of them matters from a developer point of view.)
I suspect the numbers are less horrible (but still ghastly compared to iOS) when you look only at retail sales (and by number sold, not by device count, and remembering that three or ten models of Samsung [e.g.] for different carriers might be for these purposes identical, despite their names and case designs differing) and restrict yourself to phones and tablets someone might actually be using.
Even a few dozen catch-all mostly-compatible-with-everything-you-care-about targets is a lot more than … three or four.
(Indeed, I note over there on their graph that 20 devices comprises about a quarter of the targets who downloaded their app.
Target the big, recent phones and perhaps a few big sellers from a year or two a go, and call it good and move on.
Or just develop for iOS…)
You forget about the people who develop for both. You don’t have to be exclusive.
As an Android developer, the varying screen sizes aren’t much of a problem, and there are lots of built-in features to make life easier. There are also compatibility libraries that make dealing with feature differences between API levels less of a headache, too. (And I’d rather have been dealing with Android’s layout system than iOS’s Auto Layout until relatively recently. Now that we’re headed for reasonable screen size “fragmentation” on iOS, too, things seem to be improving…)
The things that have caught me out are the undocumented, difficult-to-predict Android fragmentations, like the occasional device that has a smaller OpenGL texture size limit than you’d expect, the hard-to-handle camera hardware fragmentation, or strange adaptations of the stock software by particular manufacturers. But it’s nothing too hard to cope with well enough.
Also, it’s not like we don’t already have massively successful industries that have had to cope with insane fragmentation in the past—just look at the web! There are still people viewing pages on Windows XP with IE6 from a 17″ 4×3 CRT right now, and the next person along will be viewing it on a retina iPhone. As a web developer, or an Android developer, you just have to choose which customers you’re going to work hardest for, and where to cut off support…
how do you figure it will be screen resolution fragmentation on iOS? nothings been announced, so we have no idea what the resolutions will be and how current apps will resolve.
but yes, apple has been pushing devs to build adaptive apps for some time now.
That wasn’t a prediction about the future so much as an observation on the current state—iOS started with one resolution and size, and pretty much every year new devices have appeared to add to the number of resolutions and sizes you have to cope with. That seemed to lead to a sort of “playing catch-up” feeling compared to the Android tools and APIs, which were obviously designed for multiple different screens from the beginning.
Having said that, though, I can’t see us iOS developers having to cope with fewer resolutions or sizes by the end of this month…