The topic at December’s London Mobile Monday meeting was “Device Fragmentation and Java”. I think it’s another phrase for “diversity”, with a few “challenges”:
- Different user interfaces: different screen sizes, for example.
- Different capabilities: some devices have Bluetooth support some don’t; some devices will play some media formats, but not others; differences in performance characteristics…
- JCP specification that aren’t tight enough, leading to different interpretations of the APIs.
- Configuration issues: handsets not configured for data access, or the operators messing with data as it goes through the network, or different behaviours depending on the tariff the customer is on (apparently).
The first couple of issues aren’t going to go away, and applications are always going to have to adapt to the capabilities and the format of the device. Is that such a bad thing? A big screen and a keyboard lends itself to a different UI to a smaller touch screen. From the business perspective, though, the question is: at what point is all the testing going to kill you?
To some extend the second item, and definitely the third item, is being addressed by the MSA (JSR-248 and JSR-249). It’s all about tightening up the specs—putting in more MUSTs and less of the SHOULDs and MAYs—and making the TCKs more rigorous. That has the potential to make a substantial difference, but don’t hold your breath.
The last couple of issues will benefit from the development community pushing back on the handset manufacturers and the operators. It’s not acceptable for, say, SonyEricsson to charge me for the privilege of reporting bugs: sometimes you have to do it to bottom out an issue, but it’s not something I’m going to do unless I really have to. I’d like to see less emphasis on the self-service forums, and an opening up the bug databases. With that, though, comes the burden on developers to submit simple reproducible bug reports.