3 ms·
9 releases in 3 years -- anyone else think that's too much? I'm a mobile developer and just keeping up with Google/Apple updates is a huge task. It's almost com
by psychotik 15y ago
9 releases in 3 years -- anyone else think that's too much? I'm a mobile developer and just keeping up with Google/Apple updates is a huge task. It's almost come to a point where they ignore serviceability of products and just keep churning out new ones, expecting the eco system to keep up. And given Google's 'excellent' QA, my code is full of runtime-conditional checking to enable/disable features based on the user's version. Since users are slow to upgrade (carriers slow to push updates), the test matrix just keeps growing and keeps getting more complex.
Each SDK introduces new permissions, so now your app doesn't auto-update anymore. Which means you have a bunch of different versions of your app floating around, without converging. (I can write a whole essay about how their permissions stuff is broken - see iOS' model for a non-broken alternative).
There is a reason Honeycomb adoption is abysmal - developers haven't gotten around to customizing their apps for it, which in turn results in lower sales of Honeycomb devices. And keeping up this pace isn't going to make it easier.
Am I just being unreasonable here, or do others feel some of this pain too?
- drivebyacct2 15y agoWhat problems have you have with backward/forward compatibility with Android releases?
- psychotik 15y agoHave you tried using the Android Compatibility Library? It has several bugs, well documented with repro steps, that aren't even triaged. So now there's tons of Github forks, each with custom fixes. And when Google releases an update, all hell breaks loose. As an exercise, try converting a mildly complex app from using Activities to using Fragments. I guarantee you that you will need to sacrifice some functionality (for instance, try using ViewPager with ListFragments, and implement keyboard filtering). Now, hack stuff till you're only sort-of broken. Then, try to make it work from v1.6 to v3.0. Then, test it all in the ass-slow emulator (because you're unlikely to have devices running all those versions). It'll drive you insane, I promise.
- jarek 15y agoI thought Honeycomb was a tablet-only release. It will necessarily have a lower market share than universal/phone Android releases.
- psychotik 15y agoSure, but API concepts introduced in Honeycomb are (hopefully) being propagated through v4.0 and beyond. So apps have to catch up when they move to v4.0. There was a big push to get apps to build UIs which work on phone and tablets, which not many developers adopted because of immense number of bugs in their Compat Libraries. v4.0 is just going to make it worse -- my users are still running 2.2/2.3, so I can't abandon them. And I refuse to maintain multiple code branches. If you know the details and spend a few minutes thinking about what's involved in making stuff 'just work', you'll realize how painful it is. Google's solution is to let devs release multiple APKs per platform - that is it's own maintenance nightmare (bugs/servicing, testing, dev environments, etc).
- anigbrowl 15y agoI thought Google's solution was to move past that and have ICS run on everything, degrading higher-end functionality gracefully within the OS if it finds itself running on less capable hardware.
- psychotik 15y agoThis isn't about hardware, it's about software. Confused?
- anigbrowl 15y agoYes. 4.0 is meant to run on all Android devices, so why would you be abandoning your users if you start developing for it?
- jarek 15y agoFor the record, I read "There is a reason Honeycomb adoption is abysmal" as relating to users' adoption of Honeycomb (which seems necessarily linked to adoption of tablets, IMHO much more so than to the amount and quality of third party software available) rather than Android application developers' adoption of Honeycomb APIs (which I'm now thinking might have been your intended meaning, and which obviates my entire post).