Re: @Lee Dowling
The latest boards do not include my board. I would have to buy again. That makes early adopters, second-hand boards, and even boards from an unofficial supplier, a lottery. Ignoring that, I don't *have* power problems, because the second I noticed the specs I put it on a supply that gives a regulated 5v, 10A if necessary. But others still will because it's not about just a small tweak to fix all power problems - the board has too little voltage regulation and relies on the (not necessarily supplied or approved) adaptor to provide it. And worse than not working, it is more likely to die under load while otherwise appearing normal.
And these are not teething troubles (and if they were, my "teething" on the device was months ago now). The latest kernel does not fix the USB problems. It skirts around them by playing with interrupt rates and the SD card controller so they are not as swamped (in fact, there's little code change in terms of C code, you're just scaling back the options sent to the controller on initialisation - this "fix" was previously available as a kernel command-line option though I believe the defaults are just changed now). I, and a whole forum thread since the day of release, still have demonstrable problems with USB on simple devices. Basically, you have to sacrifice USB and SD performance to make the thing work reliably and with heavy USB bus usage you can't always sacrifice enough to make things work. And it took months of flat-out whinging to get the problem recognised and partially resolved against a background of reproducibility and wanting to fix the problem myself. The problem is not the problem, the problem was the complete lack of testing and a bad design and, up to a point, total denial and ignorance of the problem (the initial thread reporting USB problems went unanswered for months, the kernel git "issue" tracked the same, etc.). With the early adopters, testers, and people who were trying to put this device into schools and get it doing what it should do trying to help. Same with SD card problems (though that's resolved in the same way as, and was basically caused by, the same problem!). Same with training material problems.
Supply issues? Agree. Pretty much sorted. Discounting the "what version of the board have I got" question, but you can get them from all sorts of places now. Good for my casings, I'll give you that, because there are much cheaper options available now. P.S. schools, generally, shouldn't be using the devices uncased and it won't be long before some H&S fanatic picks up on that, especially if you have to have a hefty PSU to keep the power stable.
I have one, ordered before I even knew they would ever exist for real, so I was in support of the project and the ideals behind. I was basically intending to push it inside my school and get them to buy dozens, if not more, and teach "real" programming for once. But since release it's barely become nothing more than pushing hardware into schools with no thought, which is my main complaint in this thread (see: iPads in schools, too - great idea, with correct use, completely misguided to just assume schools won't just buy them and expect them to do what they read about automatically). There is no material, no training, little staff involvement, which is what I had threads about on the forums before the thing was even released. The usual answer was "we'll worry about that later", and there still hasn't been much movement in that way.
There *are* still problems. You are still going to get people plug in a USB thing and have it die mid-lesson because of power / bus bandwidth / interrupt rate problems that you can't fix (and which are down to a shoddy design of that section, but hey, that happens everywhere). It's pretending that the thing is perfect and works that gets me, and silently ignoring problems until they become major issues attracting exactly my sorts of comments on them (bad PR if nothing else).
I will be visiting BETT in the New Year. My guess is that if I even find a Raspberry Pi stall on there, it will be run by some educational supplier - because selling the hardware is one thing. There won't be training on it, which means it's dead in the water. Claiming that it's aimed at education, receiving government and media attention by claiming that, and then not doing anything to put them into schools beyond selling a board is another.
To be honest, I was prepared to buy a dozen or more and run my own lunchtime club in the school I work for - purely because I would love teaching programming. I've done it before for maths clubs in other schools and I'm not even a QT. Not a chance since I bought the device, that I still own (and still hope to get working for personal projects some day). And if I was to be given money to do so, I'd base it off something else now anyway. It's not the product but the attitude to fixes and genuine, real-world concerns about exactly how the product is pushed that worry me. Hell, when resolved the SD problems, I was basically told that the people in the UK had no idea how the drivers worked and it was only the Broadcom employee in Taiwan that would be able to hack on the SD driver to spot any differences, and that only in his spare time, and that only when granted by Broadcom. Pretty much the same answers hit for things related to the GPU ("the first completely open-source drivers" and such headlines, for an OpenGL -> RPC call wrapper? Again, that's just not true and headline-garnering at the expense of the truth).