Home > Uncategorized > Google Fragmentation Denial

Google Fragmentation Denial

Denying Fragmentation is an issue for Android isn't going to help address that real and growing issue.

Dan Morrill, Open Source & Compatibility Program Manager at Google posted a couple of days ago On Android Compatibility saying, essentially that Fragmentation is a non-issue on Android, and it has been trumped up into being an issue by bloggers/pundits looking to drive traffic.

While I am a big fan of Android, I have to respectfully disagree with Dan.  “Fragmentation” is a real issue if (1) it takes noticeably more development resources to support the multiple instantiations, and/or (2) end-users have to become hardware platform aware in knowing whether or not a particular piece of software will work on their device.

Re (1).  I don’t know which developers Dan is talking with, but the small (less than 10 developers) companies I talk with complain vociferously about the extra resources it takes to develop, test, and debug for the various Android handsets out there.  Often, it is a problem of simply getting access to the handsets before they go into the market.  Of course, one has to expect some overhead associated with supporting multiple devices, but the current level is legitimately concerning to developers–especially as they look at all the new Android devices coming to market in the foreseeable future.

Re (2).  If you want to know whether or not Fragmentation is an issue from the end-users perspective, just spend some time reading the user reviews of anything other than the top apps.  It has become fairly common practice for end users to list what handset they are using the app on, because it is widely recognized as being a key determinant as to whether or not a particular app will work.   (Dan uses the example of an app using the camera not working on a device without a camera, which is sort of a customer-intelligence-insulting example.)

I’m not saying that the Android team isn’t going to heroic lengths to attempt to avoid Fragmentation, but Denial is generally not a good approach to dealing with a real problem.  Let’s not have a repeat of the Java fragmentation tagline: “Run Once, Debug Everywhere.”