Hi Sean - there will be plenty of opportunities to link up in the near future, as my recent posts about the Computer History Museum hosting user group meetings, presentations, etc., indicate. Besides, why are we playing with a 3-D powerhouse if we can't meet in virtual reality?
Yeah, it really is painful seeing Ye Olde Same Olde Questions being asked over and over again. We need to get Eben onto the major network and newspaper news outlets to publicize the FAQ answers once and for all ... except for the cable-cutters and almost everyone under 30 who pay no attention to traditional news media.
The educational market will provide all of the numbers needed to get the Pi into the millions-fielded Hall of Fame within the next six months, if
some money can be redirected from the troughs of the ICT hardware and software bloatware vendors in the UK and everywhere else. I don't think the Foundation really understands the dimensions of that Goliath and the politics involved in keeping the vendors in the lifestyle to which they have become very accustomed. I don't care if you came up with a completely free alternative, they would still turn on their campaign fund spigots and make damned sure no one who matters (i.e., the political purse-string holders) hears a truthful thing about the Pi ever again. It's extremely difficult to thwart their PR machine, especially when demos of the Pi running CPU-intensive software slowly will be used to say, "See, this isn't a serious computer worth any amount of money." It won't matter that the Foundation will be trying to say, "No, you don't understand, the focus of the Pi is on teaching software development." As I've said before, I think it's a huge mistake to keep saying that the Pi is meant to teach programming - there's a hell of a lot more going on in any real-world software development project than programming or, more accurately, implementation, and many more people get paid for those ancillary tasks beyond just the implementation (testing comes to mind, not to mention a much-needed focus on fool-proof design principles).
It's probably already too late to get much funding redirected for significant Pi buys by educators for this coming academic year, anyway, at least beyond the early adopter educators in relatively small numbers. Even if the Computing At School initiative developing material for teaching software development was finished, it's too late to provide the teacher training for this coming year, and the unions would have a field day if anyone tries to accelerate or even skip that step. That could, in fact, stall the growth of Pi sales once the current one-each demand is met (upwards of half of which were back-up orders, as we are hearing from nearly everyone who tried to order on D-Day). The CAS effort isn't anywhere near being complete, however, and I know it wasn't meant to be, but, it's going to be yet-another speed bump that may not be visible in most Pi promoters' windshields, yet, but, like the many other unforeseen bumps and slippery surfaces that have plagued the roll-out (for whatever reasons), it's going to slow the adoption below what may be hoped. I absolutely hope I'm wrong about this, but, I've starred in this type of horror movie more than once before and, despite pleas from the movie audience, once the guy in the plastic mask starts slashing, it's just too late - the script has already been signed, sealed, and delivered. I also hope to be able to blunt some of this through influential friends in the education, business, and, most importantly, political sectors, but, this sort of thing takes way more time than reasoning people can possibly imagine.
The commercial and hobbyist markets might actually be the white knights that will come galloping in from off-stage to save the day while the educators get their ducks lined up and quacking straight. This will only be true if people can be properly educated that the ultimate trade-off on the Pi has always been cost/price whenever a decision has had to be made. Based on the questions we keep seeing, that message has clearly not been getting across even to people who should know better that you can't get something for nothing. I blame it on the generally poor overall teaching of the Three Laws of Thermodynamics, even to many engineering students these days - (Zeroth Law) matter and energy can neither be created nor destroyed and only transformed into each other, (First Law) in a perfect world the best you can do is break even, and (Second Law) the world is not anywhere near perfect so the best you can do is much less than break even ... always (aka entropy always increases over time within an entire system such as the universe, at least as far as all the evidence points, to the present).
One huge advantage of the Pi is that you don't have to keep re-establishing comms between a host and an embedded system in order to download updates to the embedded components during development, particularly rapid prototyping, which is often not rapid at all, given that infernal delay when repeatedly having to reset everything. Once integrated "Pi-duino" systems with I/O-rich multi-port serial, D/A and A/D capabilities occurs (and I know of a dozen projects currently in stealth mode kicking off in that vein alone), embedded developers will be at the intersection of Heaven, Nirvana, and Shangri-la. For commercial Pi expansion projects, it's going to be very interesting, but, see my warning in the next paragraph about expectations vs. reality.
I'm continuing to see a dangerous trend of over-hyping the Pi as yet-another Linux box (albeit cheap) when it clearly cannot be that with the current ARM CPU, and that's confusing and disappointing people when they actually get to see demos with which they can play (I'm repeatedly seeing this first-hand). The Pi can barely run a browser (or much else) at less than 100% CPU load (although I do need to try out the ongoing armhf port recently announced). I don't know if anyone has tried running any of the OpenOffice variants, but, I'm guessing it won't be a happy camper at all if 128 MB of RAM is dedicated to the GPU, as needed for 1080p full-on HD output.
The emphasis has
to be on the GPU's capabilities, and that's a problem because there are so few people with experience with OGLES 2.0 outside the iOS and, to a lesser degree, Android developers. Those communities might not even recognize how OGLES 2.0 is accessed on the Pi given their Objective C and Java bindings, respectively. It's already obvious to me that the EGL-based code needed to access the framebuffer on the Pi is so very different from that in most of the legacy demos, examples, and full application implementations which typically use completely different GL-based frameworks that probably will not fit within the Pi memory constraints. That means we're going to need to port a lot of non-OGLES 2.0 code over to the Pi, and that's going to be very painful, especially the large body of OGL code that's been building for going on 27 years (if Original Flavor GL is included, most of which as been moved to OGL or abandoned).
I'm now not even sure if the X11 GPU acceleration is really that good of an idea, but, we'll see what the three groups that are now apparently attacking the problem come up with - once again, it's that memory limitation that's going to severely limit performance, especially for the majority of things that the GPU can't help accelerate. I'm starting to come to the realization that what we need instead of X is a 3-D version of X - call it X-Cubed, or something like that. Why bother staying stuck in a 2-D desktop metaphor when half our memory and 99% of the performance of the SoC is associated with the GPU? I'm having side discussions with people outside the Pi community that have a lot of experience in standards-based (both de facto and IEEE/ANSI/etc.) 3-D environments to determine what the best candidates might be to adopt, especially given our memory constraints. I could use that for Pi-finity! in spades, as it's not clear how much of game engines like Panda3D can actually be ported to the Pi due to the memory and CPU limits.
This obviously deserves its own thread(s), so, I'll quit while I'm this far behind already
We should discuss candidate 3-D environments in such threads, and I'll be starting to look at them over the next couple of weeks, on top of the Panda3D evaluation for Pi suitability. We may need to either trim down one, or more candidates, or build something even better, knowing what we now know about user experience psychology, visualization, optimal 3-D resource allocation, etc. The future is going to be, as the Chinese blessing/curse states, "interesting".