Review: Raspberry Pi
Bare bones computing
No review is ever written in isolation, absent from context. Usually, you can guess the reviewer's bias within a few paragraphs, and compensate accordingly. He hates LCD display quality, she dislikes proprietary software, they yearn for the days of full bandwidth vinyl. You get the idea.
Raspberry Pi Alpha board – a work in progress
So here's some context up front. This July, I'll have been writing hardware and software reviews for 22 years and, for me, reviewing the Raspberry Pi (Rπ) is a first on two counts: one, I've never had to review what amounts to a physical manifesto before, and two, as a 43-year-old father of two wee boys – squarely pegged in the reliving-my-youth demographic – I so so so want this pocketmoney PC to succeed.
But understand, this is neither pure nostalgia, nor am I alone. The stated aim of the Raspberry Pi Foundation is to provide a modern, inexpensive counterpart to "the Amigas, BBC Micros, Spectrum ZX and Commodore 64 machines that people of an earlier generation learned to program on." As if you were in any doubt, even the naming of the Rπ itself tips the hat to the stalwart of that bygone era: Model A (one USB port, no Ethernet) and Model B (two USB ports, 10/100MBps Ethernet).
Rπ production model has a few component positioning changes but the concept remains intact
So what do you get for your money? The Rπ is a credit-card sized GNU/Linux computer with RCA video out, 3.5mm audio out, the appropriate aforementioned USB/Ethernet setup, HDMI out, and an SD card reader for OS and userfile storage. The SD is of course down to cost, but it's worth pointing out that using an SD for the OS offers a number of benefits; replacement media is relatively cheap; it's pretty difficult to brick an Rπ in software; image restore is just another SD card away; and pupils can take their drives home with them.
And whilst the Rπ does need to boot from an SD card, it's possible from the command line to switch post-boot so that the OS and filesystem run from an external USB drive.
Source: Paul Beech @guru
Additionally, despite no software support as yet, there's a DSI connector for hooking it up to LCD panels, and the Model B has a [CSI MIPI connector] for adding a camera down the line. There are also GPIO pinouts (though limited to 16 mA according to spec), which could make expansion and hobbyist extras a distinct possibility; projects to tie Arduinos into an Rπ, or use Arduino shilds on an Rπ are already off the drawing board.
Next page: Clock this...
Re: Linux computer?
It runs an ARM11 chip, which (confusingly) uses the ARMv6 instruction set, as opposed to the ARMv7 more commonly found in your garden variety smartphone, so support for the chips isn't that widespread. In theory it'll run any OS you can compile to run on ARM11, but you'll quickly find that that list of OSes quickly degenerates to just Linux, and once you start considering usability that list degenerates even quicker to pretty much just Debian (for now, anyway). It's quite a step from common arm chips and a far cry from your typical x86 beast, so whatever OS you want needs to be one that is supported by a broad community of active and talented developers.
Aside from Linux, RISC OS is the one that's already been mentioned, and you should have no bother running Android on the thing, but after that you're out of luck. There'll be no chance of Windows on Arm because the RPF aren't a windows partner (plus MS aren't selling licenses to end users, just OEMs per device). But really, it begs the question, given the nature of the device; a low cost, educational/hobbyist PC/embedded electronics platform, why would you want to run anything but linux on it?
And another thing...
It's pretty disappointing that a lot of comments on El Reg about the RPi show that the spirit of doing something for and by yourself - even thinking - is so degraded, while the expectation to get things handed out on plates is strong. The fact that the Pi /can/ be used as a desktop PC is amazing, not that is /should/ be used as such.
Myself, I'm planning to use them as cheap experiment platforms for a rural wireless grid, where low power is crucial (no mains on the top of a mountain) and the ability to add local intelligence to a node should be good. For example, a local node can provide a little web server that has localised info to its won area without needing to use backhaul bandwidth, remote cameras or weather stations can be set up, nodes can be used as points in e-orienteering - all kinds of things are possible. But, and this is what the "I want it on a plate" brigade will miss, we'll have to think of the ideas and make them work all by ourselves, something the Pi makes possible. We may fail, but we'll have fun finding out. And I'd rather fail having spent £30 than £300 or £3000.
Not sure how developing something like the above using something like a Pi as ONE component, albeit a crucial one, squares with commentard 1's assumption of lock-in though. If I've done 95% of the work myself, I doubt I'l lock myself in.
Come on guys, I've still got some popcorn left.