I finally got my feet wet in Windows Mobile development this month and discovered why it’s not a popular platform anymore; it’s not very enjoyable to develop on. 

To get started, I spent a day or so downloading and installing all the needed software, which is not too bad, compared to the laborious registration and installation work involved in setting up an iPhone development environment. 

However, when I tried to run a phone emulator, it popped up a window saying there’s a gwes.exe problem and that I needed to report to Microsoft. I searched the Internet for solutions, and tried some methods suggested by developers, but none worked. Instead of dealing with the emulator, I decided to just stick to downloading the image to a phone each time and test running it there. 

The next set of problems came from the ARM assembler. It complained about some assembly code that seemed to be perfect ARM9E instructions. Like usual, I posted the problem to a developer’s forum, in this case, the official Windows Mobile Developer Forum moderated by Microsoft, hoping to get a quick answer from some experienced developers. No helped showed up, but I managed to find a somewhat awkward workaround. Since I thought it might be helpful to others, I posted it as a follow up on the forum. Several days later I saw a solid “Marked As Answer by xxx, Moderator” stamp on it. 

In contrast, Apple’s iPhone Developer Forum, provides answers to some hard technical questions quickly and, it’s always a pleasure to know many other developers in the community are exchanging ideas and trying to help. To be fair, there are also lots of active developers in the Windows Mobile Forum, but my experience was still substandard. 

This all makes me look forward to possible Android development. I would be interesting to see how Google sets me up .

Intel and Nokia announced a strategic relationship back in June. To me, as both a consumer and DSP/embedded development engineer, this is fantastic news. After what Intel brought to Apple’s MAC with its processor hardware and expertise, people should expect quite a bit from mobile platforms in the future. Though the announcement is fairly old news, I have been thinking a lot lately about a couple other related embedded computing issues. 

It’s not a secret that open-source computing is a big trend. Intel and Nokia’s collaboration undoubtedly reinforced the view that open source is going to expand, including into the DSP integrated development environment (IDE) arena. ARM’s latest RVDS 4.0, Intel’s latest C++ compiler, and ADI’s Blackfin IDE, among others, all either totally abandon their earlier proprietary IDE’s, or at least support the open-source Eclipse. I was amazed to see this rapid about-face, and take it as a proof that open-source technology is knocking on all doors of the development world. 

The other question that the Intel/Nokia relationship raises is if fixed-point computing is going to be phased out of the mobile world. The two biggest advantages of using fixed-point processors are relatively low power consumptions and existing smart hardware and algorithms tailored for media processing. iPhone is using floating-point ARM architectures, while BlackBerry phones use a combination of fixed and floating-point architectures. It seems to me that the design flexibility and shortened development time provided by floating point far outweigh the various advantages of fixed-point processors, not to mention the higher resolutions provided by floating-point computation. Intel is the leader in designing the floating-point processors widely used in desktops and servers, and I believe the company will certainly continue its dominance. As Intel becomes an even bigger player in the mobile space, I would not be surprised if fixed-point computing loses substantial ground in the near future.