A great use for that old Android phone

Generations of technology quietly accrete here. While I've got a long way to go before I ever catch up with Byte's Jerry Pournelle, who in the heyday of the magazine seemed to be commissioning a new computer in Chaos Manor every month, nonetheless, in a household of just two people (one of whom is most definitely NOT a technophile), we seem to have accumulated two desktop computers, four laptops, and four Android phones, two of which are now the previous generation of technology (an HTC Desire and Desire Z, since you ask).

I could sell them, for a risible price. But they are perfectly functional little machines and the other day I found a really good use for one of them.

I've been fiddling around with home media servers for a while. Pressing an old Asus EEE into service as a slave server to a downstairs IBM Thinkpad containing my music library did work - but any time the wireless router lost power - and power cuts are quite common out where I live - it was necessary to tediously reboot the EEE to regain connectivity.

The other day, while browsing Amazon, I came across a very interesting little gadget. The BTR006 bluetooth receiver is about the size of a USB plug, and connects to any stereo line input. It has a tiny weeny little rechargeable battery built into it and - although a charger isn't supplied - works with any standard 5V micro-USB charger. And it costs under 15 quid.

So. Plug it into the bedroom stereo. Then pair one of the old Android phones with it. And then, use the marvellous ES File Explorer (free) software to connect to the downstairs laptop over wifi and then stream from the phone to the bluetooth receiver. You can control volume from the phone, select tracks, pause, etc. etc. Just remember to set the screen timeout off and then plug the phone into a charger, otherwise when the screen blanks, for some reason, it tends to stop streaming.

Downstairs, I added a cheap, long-range (30M) bluetooth adaptor to the main music streaming laptop. This allows me to remote control the main stereo. One very useful feature of this is that when learning a piece of music I can now sit at the piano in the lounge and use a phone to control the stereo. Very handy, and it works remarkably well, without any dropouts or glitches. And much cheaper than the dedicated solutions you can purchase.

The Strange Case of the Hardware problem that wasn't

During my bout of Raspberry Pi hackery yesterday, I realised that, currently lacking an HDMI to VGA converter, the only way I was likely to get anything displayed was by connecting the composite video out to something... and immediately I dug out my little Dell M109S LED projector.

Which, as it turns out, did manage, for one brief moment, to lock onto the Pi's output before giving up.... but that's life. Incidentally,  I had already had some problems getting the Pi to boot from various SD cards before I figured out the magic incantation which, to save you trouble is:-

1. Give up on the useless Win32 disk imager they recommend. It doesn't work on most laptops, you can't view the destination drives. Flashnul, the alternative, is a much better option. But I found that it was important to zero out the SD cards before use. Otherwise only one of three cards actually booted. To do this, use the -F option in flashnul. Also it's important to use the physical drive number - typically 1 - and not the logical drive name e.g F:, for all operations. BE CAREFUL!. This is a semi-automatic weapon with, to be sure, a safety catch but pointing it at the wrong drive could be catastrophic.

2. Then use the -L option to load the image and copy it to the SD card. Voila. Zeroing out seems to fix a lot of the supposed incompatibilities the Pi has with various SD cards. I have no idea why, it just worked for me.

Note to the Raspbian crew.(BTW I love the name of the distro. It still makes me chuckle. Based on Wheezy, too!. Ah, the joy of open-source names where no marketing droids tread!). Could you give us a variant distribution where the IP address is fixed e.g say 192.168.1.200 or something and VNC server is already installed. Then we could just hook up a crossover internet cable and VNC into the thing. Since I suspect 99% of purchasers already have a perfectly good PC, this is much cheaper and easier than purchasing adaptors and fiddling around with keyboards and mice, just to set this up once and then connect using VNC from then on. (which is what I intend to do, but I'm stuck by the chicken and egg scenario at present, unless I care to fire up a Linux VM and hack around with the SD card.)

But I digress. Having got the Pi to boot, and very briefly seen its initial startup menu, I realized the projector just wasn't going to sync up reliably. I turned it off and then on again just for one final check and - zap - the white background had turned cyan. Hmmmm....

Cycling power didn't help. Nor did a factory reset or any amount of fiddling with the RGB levels in the setup menu. It looked like the red LED had partially failed. Not completely - there was some red in the test patterns, but partially. So I pulled the lid off in the vain hope that reseating a cable might help, managing in the process to separate the cable retaining clip from the connector that attaches the keypad, and then spent a good ten minutes fiddling around under a strong magnifying glass trying to reattach the clip, which is the size of a sliver of fingernail, back to the connector. I have no idea how these modern gadgets are assembled. When one finger dwarfs a whole cable and connector, and resistors and capacitors are reduced to grains of sand, it really does feel at times like you're a huge and clumsy giant, something out of Gulliver's Travels.

Of course, all this proved to be pointless, but when something's long out of warranty, what is there to lose. Sadly, I added it to the long list of gadgets that had passed away in the course of service, over the years. Of course, needless to say, no such thing as a service manual exists - at least in the public domain. And I could already hear the sarcastic laughter over the phone if I were to ring Dell for out-of-warranty service.

I turned it back on and off. It was still as blue as I was. But, oddly, on power off the screen flashed momentarily white. That seemed odd. If the red LED had really had a partial failure, how was the projector managing that feat?. So I took a closer look. With no input source selected, adjusting R,G and B values had no effect at all on the screen. That seemed unusual. And where were all the other options I vaguely recalled seeing, about 'degamma' (whatever that is) and contrast and brightness.

I reconnected a VGA source. Now the extra controls were visible and had some - but not much - effect on the screen. On a whim, I wound down the blue and green settings to zero and then wound the red up to 100. It was feeble but the screen was definitely pink. Then I cycled the power again.

This time, the red seemed much brighter. To my astonishment, setting all levels back to 50 restored my normal white screen. This survived several more power cycles and seems to be back to normal now.

This is certainly one of the odder software bugs I've ever seen. Somehow the projector lost control of the LED levels, but only partially. I've never seen something that looked more like a hardware problem than this. Ah well. If you have one of these projectors and have a similar problem, I hope this helps you, before you give up on it. Dell probably know about this but if they do, there's nothing I could find on the net.

The Law of Unintended Consequences

There's no doubt that the Raspberry Pi Foundation have done great things in bringing the diminutive Pi finally to market. Mine finally arrived this week and it is amazing to see how much hi-tech componentry is crammed onto a tiny board for a risible sum of money.

But the Foundation's aim - to introduce a new generation to computer programming - seems oddly off-target. Sure, they're selling bushels of these things. But apart from some canny school procurements  - motivated mainly, I suspect, by opportunism - you can loosen the purse-strings of wealthy IT-literate sponsors by appealing to their early days spent tinkering with technology - most sales, I think, are to people who have a great idea which needs a low-cost embedded controller to make it happen. And make no mistake, the Pi is an amazing bargain compared to just about any other solution on the market. You'd be very lucky to find anything at five times the cost with comparable power and connectivity.

But let's come back to that grand plan to engage the youth of today with the joys of programming, tinkering directly with the hardware. To make that happen you need not just a Pi, but an SD card, a power supply, a keyboard and mouse, a monitor and - due to the odd decision to go with HDMI out rather than VGA - most probably an HDMI to VGA active convertor as well.

By the time you've gotten this together the costs are likely to be comparable if not higher than those incurred just going down to the local charity shop/pawn shop etc and picking up a perfectly serviceable laptop or desktop that is several years old but still perfectly capable - a Pentium M-powered laptop, say, or an Athlon-powered desktop, complete with hard drive, keyboard, mouse and monitor. From which you could learn programming perfectly well.

But this would hardly be an objection if the Pi - as its creators imagine - is really an homage to the BBC Micro. Such a machine would be intimately 'knowable', down to the merest register and I/O pin. Certainly the early tinkerers knew their machines completely. The achievement of such mastery is what led so many people into a lifetime working in IT.

Sadly, a Faustian bargain was made in bringing the Pi to market. The creators wanted something powerful enough to handle real-time video, and so they opted for a Broadcom SOC which included a GPU. In so doing they accepted that Broadcom would not release any technical data on the GPU and that all communication with it would be by means of a binary 'blob' of code.

So let's think for a minute about what lessons a young would-be programmer might learn from the Raspberry Pi. Firstly, they'll learn that curiosity has limits. Portions of this device are firmly 'off-limits' to the enquiring mind. Should you be curious about how the GPU might be used for massively parallel programming, or wish to experiment with video capture (perhaps image recognition) and processing, well, you can forget about that. The information on how to do this is only available to a privileged few.  Want to connect a camera?. They're working on an 'official' one - but will you be allowed to see all the source code for it?.

And let's suppose for a moment that an enterprising young genius - another George Hotz, shall we say - decides to try reverse-engineering that binary blob and publishing details of its inner workings. It's not hard to imagine Broadcom coming down on them like a ton of bricks. This isn't going to be good publicity either for the Foundation or Broadcom.

Now, these issues have been raised on the Raspberry Pi forums. And the foundation staff get quite defensive about this. It's understandable that they felt this was a reasonable compromise, given their misguided idea of what might attract a fresh generation of programmers. But giving them a wonderful machine with a locked room and a big 'keep out' sign is not a great idea at all. Instead, it teaches those young minds that information isn't freely available to all, and that they can just, like Dickensian children, stand at the window and flatten their noses against the glass and stare at the unobtainable things on the other side.

And another thing they'll learn from the Pi is that, thanks to the wonder of software patents, their great ideas might not belong to them. No, they could write a marvellous piece of code and then find that they have to pay someone else a fee for the privilege of doing so. (that's why the Pi has no software for video capture, for instance. They have to pay licensing fees for it, unless they use the royalty-free OGG format).

Another lesson they'll learn is that under every bridge lurks a bottom-feeding patent troll. Who'll claim that their patent on 'clicking a mouse to display information on a video display device' covers that cool new piece of software our youngster just wrote and sues them for ten million bucks, forcing their parents to sell their house just to cover the legal fees.

And that's a shame. Because among those great young minds are tomorrow's geniuses. Who will turn their talents to other fields - physics, mathematics, chemistry, and so on, where knowledge is not locked up in secret vaults but made available to all, for the common good.

I can only hope that at some point Broadcom might see the logic of opening up the inner workings of their GPU. Imagine the amazing things that might be developed, if only we - and our children - could be given the key to the knowledge locked away for corporate gain. Meanwhile the Pi remains a flawed gift. A great gift - and we should be very grateful for the Foundation's hard work - but, regrettably, not the gift that keeps on giving.

Secret Herbs and Spices

So you want to learn more about how your PC actually works?. Perhaps you'd like to know more about the inside workings of your Intel or AMD processor. Well, that's no problem, of course. Hugely detailed technical datasheets are available from both these vendors. Sure, they are a bit reticent about some details - internal debugging registers, that sort of thing. But, for the most part, if you want to know what connects to each of the pins, what the instruction sets and registers do, and how the device works internally, all that information is available.

And so it is for most electronic devices. After all, how could engineers design without this information?. Or hobbyists learn the secrets of circuit design?.

Unless, of course, we're talking mobile device CPUs (and GPUs). Suddenly it's a world where, unless you are (as Nvidia state in their developer forum) 'considering a market for 100,000+ devices', then frankly, you're out in the cold. For some reason most of the mobile CPU vendors are incredibly secretive about the internal workings of their ARM-based SoC (System on Chip) devices.

Oddly, Texas Instruments appears to be the odd one out. They do make detailed technical reference guides available for their OMAP series of devices. I'm not quite clear what earth-shattering secrets the other vendors feel they can't disclose, but for open source developers this paranoid secrecy must be absolutely infuriating.

But if your device comes from Qualcomm or NVidia, forget it. Their device secrets are available only to a select few.

You can understand, perhaps, that these vendors hardly want to support hobbyists who want to tinker with their chips. But TI's relatively open approach has made it possible for a whole open source infrastructure to spring up around their chipsets, such as the Beagleboard. I think the other vendors are really missing an opportunity here. If they persist in their paranoia, in the long run, it's likely that the more open vendors will get the attention of the Open Sourcerors. Already, Canonical have made a real effort to support the ARM architecture, for example, but these attempts will founder on the proprietary extensions made to the architecture by each vendor, particularly if the technical details have to be reverse-engineered.

So, Qualcomm and NVidia. This is a plea to open up more. What secret herbs and spices do you really think your competitors don't know about, that are buried in your private datasheets?. Open up and let the innovators come knocking at your door. It's your market to lose, long term, if you don't, regardless of how many devices Samsung or HTC are buying off you right now. When the market matures you need to look for diversity, and that comes from smaller players, who may be considering integrating ARM and Android into all sorts of products, such as car dashboards, refrigerators, stereos.... who knows what. If they can't easily get technical data when they're doing a proof of concept, they'll go to the vendors who are willing to engage, and design around their products.

You can see this quite clearly in the embedded processor market, where Microchip made the PIC series of processors incredibly popular by actively encouraging hobbyist designers. No matter how technically amazing your products are, its a maturing market. Imagine if Intel and AMD kept their processor datasheets secret. Would we even have Linux or Android now?. I doubt it very much.

Corporate denial

A few months ago I purchased an Archos 101 tablet. I was very impressed with this device - sure, the screen viewing angle isn't as good as the iPad, there's only 256M RAM and you need to install a hack to use the Google Market, but at half the weight of an iPad, its a nice little package.

But hardware engineering ain't always easy. Apple learned that with the iPhone 4 antenna fiasco, and Archos have learned the hard way with the A101.

Specifically, there's a problem with the USB host port. It will fail. When it does, the device will also stop being able to sleep, so you're up for turning it off when you're done using it and then rebooting every time you need to use it again. Otherwise the battery will run flat in a few hours.

Of course, I didn't know there was an endemic problem with the device until I had the problem. Now, modern electronic devices are so reliable (six sigma and all that), that it's been a while since I had a hardware failure. And pretty much every hardware failure I've had in the last decade can be traced back to the humble capacitor. A lot of these have failed due to a fascinating story in industrial espionage gone wrong - but you can google 'bad caps' to learn more about that one.

This was different. And, regrettably, so was the way Archos seem to be handling the situation.

Now, before I continue, let me just say that Micro Anvika, from whom I purchased the unit, have been very professional. I took it in, they accepted it for repair and what happened next was entirely out of their control.

What happened next was..... nothing..... Archos accepted the unit for repair and...... nothing. Nearly six weeks later it was clear that getting things sorted in a timely manner didn't look likely. Micro Anvika were quite willing to replace the unit, but by then I'd read all the postings where others had apparently had very similar issues. So I declined, and got a refund.

The point of this, though, is not that things fail. I accept that. Recently I had an LG monitor fail. Bad caps, I think. I was all ready to go and buy another one when I checked the warranty and found to my astonishment that LG warrant the monitor for a full three years. It was two and a half years in, so I rang them. Yes, sir. Bring it in to your place of work and we'll have a courier deliver you a replacement. Which they did. Fantastic service.

Archos, though, appear to be in denial. Is there a serious problem with the A101?. Well, forum posters think so. But Archos remain silent. The problem here is that this silence, coupled with an apparent inability to turn around warranty repairs in a timely manner, really impacts consumer confidence. I'd have accepted a replacement if I'd been confident in the company, but clearly, they aren't providing an acceptable level of after-sales service.

Companies have to learn that burying bad news always makes things worse. Right now, I'm very wary of ever purchasing - let alone recommending - any product from this company. Yet if they had been open about the issue and resolved it promptly, I'd have been more than happy to give them another chance.

I will be writing to the CEO of Archos about this. Reputational damage is very hard to overcome, and it's a shame - their products are innovative and cost-effective. But consumers are much more informed than they were a few years ago, and negative press will quickly pile up and overwhelm a company which buries its head in the sand when problems arise.

A rude shock for HTC fanboys

So the original HTC Desire isn't getting Gingerbread after all. Not enough RAM, apparently. You'd certainly think 576M was a fair amount, but obviously not if you want to run Sense.

This is a blow for Android, because clearly pretty much all the devices, regardless of vendor, released prior to 2.3 will probably now never get an upgrade. Sony, Samsung and the rest will breathe a sigh of relief; if HTC weaselled out, so can they. And they will.

Of course you can root the phone and install Cyanogen, and I'm wondering how long it'll be before some of the more enterprising 'mom and pop' mobile phone stores start offering this service, managing the tedious task of getting all your apps and data across, for some kind of (hopefully not too steep) fee.

It's also a blow for vendor (and carrier) customisation, because if this is seen by customers as potentially blocking future upgrades - and it certainly has had a huge impact in the timeliness of upgrades in the past - there will be growing pressure for handsets to be offered with stock Android builds, further commoditising a market already rapidly heading the way of the PC.

But it's also a worry when it comes to malware. As I said a long time ago, Android's openness is a two-edged sword. Coupled with an infrastructure of devices that cannot be updated, it's a crook's paradise, since existing security flaws will be there indefinitely. Sure, the Android Market *might* catch some of the bad guys but given that software can dynamically load other components, and that the kernel exploits may never be patched, how easy will it be to keep control of this.

HTC need to urgently re-think this decision. It might be a loss of face to abandon Sense - but the wholesale abandonment of its existing user base to what can only become an escalation of hackery, is a disaster in the making. How long will it be before customers sue HTC because they were hacked, for example?. Does HTC have a duty of care?. It's a tough call, given that Microsoft, for instance, have drawn a line on future Windows XP patches - but with device lifecycles as short as they are, how willing will customers be to upgrade. And even if they do, those 'non-upgradeable' handsets will make their way down the food chain. What happens when some litigious customer gets a multi-thousand dollar phone bill due to hacking and sues the carrier and HTC for negligence. Or will the carriers start barring old phones as a pre-emptive measure to avoid these kind of scenarios. That'll be a popular move!.

How ironic it would be if CyanogenMod actually offered *better* security by rooting the phone then vendor locked-down customised Androids did.

Google's stewardship of Android has been pretty erratic. The lockdown on Honeycomb certainly has helped, when it comes to the market being flooded with knockoff iPad wannabees, mass-produced from inferior components, but this can't last forever. Google really needs to think about parental responsibility here. Right now, baby Android is hanging around with some pretty scuzzy lowlifes, and Mom's looking the other way. And as Android users, not all of us enjoy playing in traffic. Time to wake up, Google, pull down a curfew, ground the kid for a while and get into some Tough Love. Otherwise, it'll end in tears....

Yahboo!

I've had a yahoo business mail account for a long time now - and it's worked fairly well most of the time. Mailbox search failed at one point, but other than that, well, it's email, what can you say.

Stupidly, I assumed that because I was forking out something like 6 pounds a month for the service (and a further fee for my own domain), that this somehow entitled me to any form of support whatsoever when things went wrong. Because of course, I could have had my domain hosted by gmail for free.

Well, I've learned my lesson now. Things started going wrong about a week ago when suddenly my phone indicated a login failure checking email. I thought nothing of it until I logged on from my home PC to be informed that my account 'might have been compromised'. Since my password is reasonably non-obvious, that was worrying, but things rapidly went downhill from there.

I tried resetting my password, since now my normal password no longer worked, but apparently the alternate email I'd set up was a hotmail account which I couldn't recall setting up. That's my fault, of course - moral number 1, don't forget your alternate email account. But we are talking something like a decade ago when this was set up.

I tried to recover access to my hotmail account but to be honest I wasn't entirely sure what the heck it was called. At any event, this proved fruitless so I tried contacting Yahoo support by email. 48 hours later I got a response apologising for the delay and telling me to ring customer support. I'd already tried that, of course, but it was obvious that this was an exercise in futility. There's an option for Yahoo small business - when you take it, an infuriatingly upbeat female voice smugly announces that telephone support is no longer available and I should sign in for support. But I can't sign in!. Ah, Kafka, where are you now. How you'd love this!. As for trying to contact account support, forget it. You just end up in a queue from hell. And, ringing long distance from London?. I don't think so.....

Meanwhile, the whole yahoo email infrastructure appears to have collapsed. Hence, it's even more impossible to get access now than it was a day or two ago. Apparently something has gone horribly wrong switching people over from 'classic' email to the new-look email. This wasn't my problem but now I am swamped in a huge queue of angry people who also can't reach their yahoo email, but for a completely different reason.

Well, I'm giving up for a few days. Fortunately, access to mail isn't life-threatening for me, but clearly it's essential to regain control of my domain and move everything to a company where customer service actually means something. A lesson learned, fortunately not the hard way, but if I had been running my own business, this could have been catastrophic.