Dear Mr Purdy,
this is an open letter and it will be posted in my blog. Allow me to start abruptly and say that your article does not make any sense. From one side you are saying that Java is not good enough because “phones are too different”, but you then say that Linux will fix this? How? There are SIX different Linux mobile-phone implementations so far. NONE of these 6 implementations are compatible! When a developer is going to create a graphical application to run on a Linux phone (e.g. a Motorola/Montavista phone), his application will NOT work without full porting to another Linux phone (e.g. a PalmSource/Access phone). Only the kernel is the same and some system libraries (and not necessarily binary compatible either), but the graphics on top are completely different and so developers must program using different toolkits all the time. In most cases this will be as laborious as porting a graphical application from Windows Mobile to PalmOS. Sure, porting Apache or recompiling wget will probably be easier, but cellphone consumers don’t care about server or command line software.
You write: “Economically, I believe all of the current mobile OS vendors will come to realize that the value in the total offering is in the services and applications not in the OS kernel and, as a result, will shift their offerings to become value added suppliers to the Linux ecosystem.”
While I don’t necessarily argue that Linux will become the dominant mobile OS, I do argue that the result will be homogeneous. It won’t. If we assume that 90% of all smartphones in 2015 are Linux-based, this would mean something *like* this: Motorola might have a 25%, Access a 20%, Trolltech a 10%, DoCoMo Linux another 5%, Mizi/Samsung Linux another 25%, and OSDL’s version about 5%. So, how is this any better from what the fragmented mobile OS market that we got today? It’s not! These systems are NOT compatible with each other! They don’t want to be compatible in fact! Each one of these Linux mobile entities thinks that they are doing the right thing and they shrug off the rest! Things are WORSE than it is today about Linux distros (which are semi-compatible) because in the mobile world, they will be fully binary incompatible. And for most apps they will be source incompatible too (at least desktop distros today are source-compatible most of the time).
So, can you please explain to me how Linux is better than Java for applications? If there was ONE implementation of GNU/Linux for phones used by different vendors, I would say, “cool, I can’t wait for such a phone”. But this is not the case because every one of these companies took Linux part by part and each are building their own car, instead of working together to build THE car.
And btw, your statement that “it seems plausible that UIQ should operate on top of Linux rather than the Symbian core to be more efficient” is way off. The Symbian kernel can’t do as well as Linux does in some areas, but this does not mean that is less efficient than Linux! Symbian was developed and architected to run well under mobile devices, albeit under-featured compared to desktop kernels. Linux was developed to be a bloody server OS that was later scaled down using a hammer and a fat lady that sat on top of it in order to make it fit on a phone. I own 3 Motorola Linux phones btw (I can send you my review URLs if you are interested): each one is at 312 Mhz, 64 MBs of RAM. And each one is considerably slower than my Windows Mobile phone at 200 Mhz and my Symbian/S60 3.0 at 220 Mhz. How many smartphone platforms have you actually tried and tested yourself?
Obviously Mr Purdy, you don’t have the knowledge required to make such assessments. You assume that companies that use Linux will create phones that are compatible with each other, while this is not the case AT ALL. I expect you correct your article ASAP because it’s misleading — or wishful thinking at best.
Rgds,
Eugenia