Here is an old article I’ve written when I was a student about software metrics/efforts… As I need for my day to day job to have rough estimates of software projects at an early stage, I’ve begun to train myself on it and resurected this old essay.
Comments rss
TrackbackThis great post asks the good question, “does mobile OS matters?”.
A solid, robust OS is only a tool to achieve “function”: ok it helps to achieve that, but do you know the OS in your iPod/Blackberry/phone/car etc? Having a big OS HAS drawbacks, like CPU overload, memory consumption … for nothing related to the device “function”. The only real advantage is that an OS offers a common developpement framework, that is a clear abstraction to the hardware and allows easy developement for a wide range of devices … but after all is that the role of an OS? Take a look at Java: it offers such a development abstraction, and relegates the underlying OS to a mere commodity (hum, … ok sort of), it also has a BIG drawback : this hardware abstraction is too strong/big/too generic for mobile devices (performance issues, used RAM and above all access to device hardware functions). After all, what is “really” needed for mobile device?
To satisfy all, you need to make compromises (like market segmentation in function and cost).
So coming back with my point of interest: Mobile phone industry. Open Plug has been founded around this idea: “only Developpement tools and APIs matters”, and above all, as said in the article “only device function matters”.
…and this industry (devices, telepony…mobility) is so big that the today’s big player won’t allow a complete industry stole ala microsoft: the big guys there are Cell phone operators, phone manufacturer, content providers…
Nowadays software developement became the undisputed bottelneck (along with mass production) when building a cellphone. To help with that (and take the lion share of a billion device market), software houses are pushing high level OS adoption in cellphones: see efforts from microsoft, symbian, palm …. anyway it adds software and hardware cost (bigger CPU, even adding CPUs, more RAM, bigger battery) and in many case it really lowers ease of use, and the first rule is many times sacrificied.
Linux may be a great free alternative BUT Linux alone is only a kernel, you can’t do any application cause there is no FRAMEWORK on top of it: no sound, graphic, telephony handling, etc. You definitively need something: Motorolla has tried to do that, Trolltech (Qt/Qtopia) is doing that, and more recently PalmSource is doing that. This last, surprising move, really shows that the main PalmOS advantage was its APIs on top of which many applications have been written, on top of which you could really make a mobile phone with coooool features (see my treo650, ok there cost is fairly high). The underlying OS (who know the kada kernel???) was really a commodity, they even tried to make a new kernel but ultimatelly failed before selecting Linux: the kernel war is over, the big players are there, too late to compete.
Here at Open-Plug we have a similar approach than Palm: building a framework on top of …. a subset of the linux API and not really on Linux itself, so all the framework may run on Linux (for High-End Phones) or, porting this Linux API subset on top of low-end/ proprietary OS/platfrom. The whole thing, coupled with an efficent component based technology and powerfull development tools may seems very similar to java in fact … but in that case:
)Sounds too good to be true ? Hey, ok I’m biaised and enthusiatic about what we are building here, but we are making progress and you will see our phones soon .
Comments rss Trackback
