Whistles & Clicks

This squalid backwater is where I put things as I think them. The more I think the less I post. Time will tell. I can also, occasionally, be found on Twitter and Google+.

This is very much not a blog. I had one of those many years ago and it was terrible. This may also be terrible, but early signs are better than that, and either way... it lacks most of the features people expect from a blog.

04 November 2015

Apparently only one of my monitors is compatible with my new sunglasses. Super. Now I need new monitors.

24 Feburary 2015

This is the most astonishing painting that you or I have ever seen, and I say that knowing that someone of your experience must have seen some pretty astonishing things.

24 Feburary 2015

I'm an ape-descended life form so amazingly primitive that I still think digital watches are a pretty neat idea. This one doubly so. If you're deeply into the google now / android notifications koolaid then you really need to get one. Every use is a delight. And every hopeless screen layout is a joy.

21 December 2013

We're a family which is very poor at keeping in touch with family and friends. But one of the rare things we do manage to organise is a set of photos for the Christmas cards. They seem to be well recieved, so here are this year's in all their cute glory.

04 October 2013

I am in blood, Stepped in so far that, should I wade no more, Returning were as tedious as go o'er.

Is there a point at which the body gives up on digestion, absorbtion and process? Once a binge has dug certain depths is there nothing to be gained by stopping? I could always do some research.

19 June 2013

I hate changing ISP. And yet I'm doing it again. Sad face.

Currently I'm with Plusnet on their fibre product. It works great (no down time since it first sync'd) and I get decent speeds (for the UK at any rate). The problem is that I have a 120GB per month cap on my line, and if I exceed that the cost rapidly becomes absurd. I've dones this once, when I left an rsync running at the wrong time, and it was a really annoying way to waste a plie of money. The line is unmetered overnight (Midnight - 8am) so I end up having to plan my transfers very carefully. Backups and server images all have to copied around via crontabs, and then killed if they threaten to overrun. This does work, I manage it with a bunch of scripts, but its annoying.

A Wikinut DB dump, for example, is 20GB even with all the cruft cut out before the copy. 20GB is the stuff you're actually likely to want to analyse. If I want a full copy for some test or other then its even bigger. So I lean on these scripts pretty constantly. Oddly this has revealed a flaw in Plusnet's accounting, they don't seem to display usuage over 1TB correctly... and I've gone over that on several months. I'm as childish as anyone, so seeing my usuage overflow on the stats is acutally kind of fun;)

I should point out that this is real traffic, for want of a better criteria. This is backups and datasets, not torrents and warez. There isn't really anything here I can avoid doing. To give Plusnet some credit I've never noticed any rate limiting or slow down. Given that much of my usage is over ssh I would be disappointed if they did, but credit where credit is due.

But the annoyance is just that bit too much. Today, yet again, I find my self having to wait until tomorrow for some data. The server is too near its IO limits to do my analysis on, and the data file is 10GB and too big to copy. My billing month ends on the 26th and I'm too close to my cap. So I cron it and wait. Silly.

I've placed an order with Sky for a very similar, but uncapped, service. The price is more or less identical with the bonus of a £10 per month discount for the first 6 months. We shall see if it's any good.... and whether they can get me transfered without downtime. When I moved to Plusnet (from Sky as it happens!) I ended up using my phone's 3G connection for 4 days. That wasn't ideal ;)

17 June 2013

  • 500g 20% fat fine mince beef
  • 100g diced smoked pancetta
  • 1 diced onion
  • 2 egg yolks (or 1 large)
  • 2 Matzo crackers
  • grated parmesan
  • coriander
  • salt & pepper

Fry the pancetta over a low heat until crisp. Set aside. Soften onion in the pancetta fat, set aside. Crush matzos into a powder and combine with the beef and egg yolk. Add chopped coriander and grated parmesan according to taste. Add cooled pancetta and onion. Season. Form into 6 patties. Cook over a hot grill until crisp on outside and cooked through. Serve with corn, portobello mushroom and chipotle hot sauce.

17 June 2013

I've been looking at a lot of colour photos from the second world war recently. There are some that really jump off the screen at you, so vibrant and sharp compared to the grainy black and white photos you more commonly see. When this happens I invariably find that the film used was Kodachrome. Arrh, glorious Kodachrome. You'll be missed. Can't believe we replaced you with Instagram filters. Such errors as we make.

11 June 2013

I'm always slightly bemused when adults at theme parks get all excited about the character meet and greets. It's not like any properly adjusted person should get excited about meeting a bloke in a mouse suit. And then this went and happened, and my world was realigned...

16 May 2013

New toys today, and what marvellous toys they are. First up is the nook simple touch I ordered a while ago at the ridiculous price of £29. It was shipped UPS from Belgium of all places. So yeah, B&N didn't make any fat profits off of that. But I haven't even turned it on, because my new Raspberry Pi CSI Camera Module has arrived too!

Getting up and running took mere seconds, and the quality of the output seems decent. Nothing amazing, but totally acceptable for a sub £20 component. I'm less impressed with the sample camera applications, raspivid and raspistill. They're fragile (segfaulting on seemingly sensible options) and do the bare minimum. This would be OK ("its the UNIX way") if they were just a little bit better. But I'm sure that'll improve. Ideally a V4L2 driver will be available at some point, which would make some things a lot easier.

There is a suggestion that lower resolutions at higher framerates (90, maybe 120) is coming soon. All in all its a good buy if you already have the pi to run it off.

15 May 2013

People - please stop making devices that output 5V at 500mA. You're building products that don't work with the rest of the market.

I get it. The USB 2.0 Spec thought 500mA was enough for everyone. No one is going to set their cable on fire by running 500mA over it. But the real world has moved on. Most phones ship with chargers that output 1A, and if you only give them half of that they take aeons to charge. Aint nobody got time for that.

I even have devices (like the Raspberry Pi Model B) that aren't stable at 500mA. Apple ship an iPad charger that outputs 2.1A! Over four times the level the specification would have you implement.

I was seriously considering buying one of these so I could run my phone's GPS all day. But it only outputs 500mA, despite having an LiPo battery between your device and the generator. Depending on exactly what you're doing with it my phone will drain its battery when charging at 500mA, whereas at 1A it'll run forever. I've been caught out by this before, I bought a cheap car charger and discovered that the phone would lose charge when running as a Sat Nav despite being connected to a relatively huge lead-acid power source.

So yeah, anything USB that outputs less than 1A is dead to me. In fact I'm going to simply refuse to buy anything that can't output 2.1A (which seems to be the emerging de facto standard now Apple have endorsed it). These deranged maniacs with their 500mA fetishes don't need the encouragement my cash would surely give them.

If my Kickstarter USB Power itch doesn't settle down soon then I'll probably buy one of these. It can probably run my phone for about 18 hours and seems a more versatile solution. Plus it can output 2.1A, which is the exact amount sensible people expect it to provide ;)

06 May 2013

I am a highly committed fair weather cyclist. Which is to say that I'm an absurdly keen fan of good weather and a reasonable cycler, which averages out. My bike is an inexpensive model that has served me well for about 4 years now, coping admirably with several crashes. It isn't very light and the shifting is clunky and imprecise. But hey, it survived those crashes and a couple of them were potential frame killers. So I can't complain. Plus I've pimped it with some totally excessive tape and tyres, the kind of foolery you'd be reluctant to force on a classier model.

I normally start riding in earnest some time in April, getting up to a few rides a week with one decent length one (50+ miles) by mid May. This year is a late start. Partly because of the weather, and partly because I'm on very low calories and bonking a long way from home is no fun at all. Today is a public holiday, and sunny, and my daughter wants to go out. I can't resist that for a combo.

A couple of hours of cleaning, fettling, oiling and basking in the sun ... and we're ready for the off. Lets hope its a dry year for all us fair weather cyclists!

02 May 2013

B&N are selling off the Nook Simple Touch for £29, including VAT and delivery. Something of a bargain for an 800 MHz TI OMAP 3621 with integrated e-ink display, decent RAM / storage, wifi, and battery.

I've ordered one without particularly knowing what to use it for... Thanks to good ePUB support and no silly drm lock-downs I may even end up reading books on it! Gaining root looks to be an easy process, and despite the ancient Android 2.1 image (Eclair based on Linux kernel 2.6.29) it should be possible to get most things running on it easily enough.

01 May 2013

Today is day 21 of the VLCD, which means I'm about to start the 2nd half of the journey. Over the hump. Home stretch. The fat lady is doing her sound check. Harry is readying the badger parade. A time for much merriment for sure.

I've been thinking about some of the arguments against VLCDs. I've always tended to pick holes in things I force myself into, but in this case the holes aren't deep and at the end of it I'm more sure than ever that I'm doing the right thing. The most compelling argument that I could come up with is that VLCDs don't teach you anything, they don't change your behaviour, and hence you're doomed to repeat past mistakes. There is some evidence to back this up, various commentators claiming that high percentages of VLCD users regain all the weight (or more) within a year or two. Worse still - YoYoing weight may be more detrimental to your health than being consistently obese.

First off - I haven't put any weight back on, and I'm about 18 months down the road. An anecdote doesn't make evidence though, but frankly I couldn't find much decent evidence either way. Yes, I could find lots of studies that showed you were more likely to gain weight in the year after a diet than in the year after not dieting. But that isn't an argument against dieting. And I couldn't find anything to show the relapse rate of VLCDs to be any worse that other diets which had resulted in similar magnitudes of weight loss.

Secondly - VLCDs do teach you important lessons. I agree that you aren't learning a new set of meals which you can live off long term, but you are learning wider and more general skills. I've learnt to be content despite hunger, something that most first world adults simply can't do. If I'm hungry and aren't due to eat for 3 hours I just get on with life until those 3 hours are up. Hunger needed be any more compelling than the desire to watch a new film. You don't skip work to see the 10am screening of the latest release, you wait until a showing that fits your schedule (even if the desire to see the film pops into your head repeatedly during the day). Hunger is the same. Modern lifestyles have given hunger way too much cultural importance, something that has to be acted on immediately. This is a falsehood. The important hunger, the one most of us never experience, is something different. It comes on far later and first world adults will probably never experience it. Once I learnt that hunger is just a data report from the body, to be noted and dealt with in due course, I was well on my way to mastering consumption management.

Another skill VLCDs give you is an acceptance of monotony. Modern man has a staggeringly varied diet, far more so than at just about any point in our history. In fact once you get past the recent couple of millennia (a mere blip) humans ate the same few foods every single day. There was seasonal variation for sure, but the staple food of a community didn't change. You ate it day in, day out, for your entire life. I'm not suggesting that I'm going to take up a diet of purely maize or wheat, but there's no harm in expecting the daily meals to be simple. Constantly striving for new foods, better flavours, or more exciting choices, leads to the spiralling addition of calories, cost and complexity. And the more food costs in terms of money and prep time the more important it becomes in your life. This makes you expect more from it in return, which feeds the cycle. Going back to simple/wholesome food based off vegetables and resets your expectations. And forcing several weeks of really basic food can reveal hidden pleasures, you have no idea how strong/sweet the taste of a raw carrot is until you avoid all processed foods for a couple of months.

I'm not sure who I'm trying to convince, clearly a large part of this is aimed at my own evangelising self. I'm totally committed to this 6 week run, and I really do feel like this is the right way to achive my goals... but even an eager practitioner like myself has benefits from a little pep talk when settling in for the final stretch ;)

22 April 2013

I've had problems correlating my food intake with my body weight. This has been the case for most of my life, certainly all of my adult life. A couple of years ago a blood test showed that I had also developed high blood sugar, leading to me being diagnosed with type 2 diabetes. As part of the investigation into that condition it became apparent that I had an atypical insulin response, and this was probably partly to blame for my problems understanding my weight gain. In short the standard advice of "lower consumption, low fat and healthy whole-food carbohydrates" doesn't work very well for me, as I jump into a fat deposit mode very quickly in response to even small amounts of carbohydrates.

Jumping to the core of the issue - I need to be eating things that are digested more slowly and release less sugars. All starchy foods (potatoes, bread, pasta, rice) are out, replaced with green veg. Its not complicated and its easy to stick to once you get used to it. I'm allowed (limited) amounts of fats and meats so satiety isn't really a problem. Using these rather basic rules I've been able to completely control my blood sugar without the use of any drugs, which is the best way to manage diabetes if you expect to be living with it for many years.

I also started to lose weight. Which was interesting, because I wasn't calorie counting and wasn't focusing on weight at the beginning - blood sugar was the main measurement I was working on. Experimentation and blood testing after every meal made it pretty clear - if I avoided "simple" carbohydrates entirely (not even a single bite!) I would both lower my blood sugar and my weight. So I did that for a couple of months and lost more weight than I'd done on any diet I'd tried in the last 10 years.

Once I had a handle on my blood sugar I became somewhat obsessive about control. I read books, countless web pages and plenty of research papers. My first conclusion was that the NHS blood sugar targets and testing regime are hopelessly lax. I set my own targets which are considerably lower and test at least daily (the NHS bizarrely thinks 3-monthly is adequate).

Even with good responses to my food intake it was clear that the best way to bring my post-fasting blood sugar down to my target of 4.5 mmol/L was to help my body by losing a lot more weight.

There is some evidence that the way you lose the weight matters for your insulin response. Really rapid weight loss (borderline "starvation" weight loss) can better promote fat loss from around the internal organs. This visceral fat is particularly damaging, especially fat around the liver and pancreas. Unfortunately visceral fat is very hard to monitor without access to an MRI machine, I was going to have to guess. So I made the fairly safe assumption that I had visceral fat problems, and the slightly less safe assumption that a very low calorie diet would shift it.

A VLCD is defined as a diet providing less than 800 kcal per day. Compare this to the NHS's recommended calorie intake for a man of 2,500 kcal per day. The plan I opted for was based on 3 commercially prepared meal-replacement-shakes (3*200 = 600 kcal) and 2 modest servings of green vegetables (200 kcal, broccoli or cabbage mainly - so many brassica oleracea to go at). Its important to use VLCD shakes and not general diet shakes, as you need them to provide everything you need to maintain your health and slim-fast type products don't do that. In fact I took a couple of supplements just to be safe, and added some salt to my veg as nothing else was really providing that.

I decided to do the VLCD for 8 weeks.

At times it was hell. At times it was easy. But throughout I lost weight at a truly shocking rate, an average of 300g per day across the 8 weeks. At some point I intend to write up some thoughts on how it actually felt. The side-effects (and mental effects) are interesting to say the least. But for now, lets just say that I made it to my target of 8 weeks and lost approx 17 kg (37lbs) in the process. I had read up on techniques for avoiding lean mass loss (HIIT seems to be the best option, but the studies are small) and did pretty well in that regard. Since the end of the 8 weeks my post-fasting blood glucose has basically been pegged at 4 to 4.5 mmol/L, which is great. I even "passed" a Glucose tolerance test. If I'd originally presented as I do now then I wouldn't have been diagnosed.

Which isn't to say I'm cured. Diabetes is a chronic condition, you can never be cured. I just have it under control.

My plan had always been to try and stabilise after the VLCD and hold my weight (and diet regime) steady for 12 months. Which I have done - 1 year after the VLCD my weight is 900g lower than it was at the end of it. Despite achieving this (and it is something to be pleased with, a huge percentage of people trying a VLCD yo-yo the weight back on within a year or two) I've decided to push on and have another crack at it. I have a much higher risk of lean mass loss now because I'm starting from a better place (and I've been hitting the gym a fair bit over the last year) but I'm pretty much a convert to the short/sharp VLCD way of thinking. This time I'm going for 6 weeks, which seems a sensible length.

Today is day 12, and I feel pretty good. This post was intended as a general background to what I'm doing. I hope to add some specific comments on progress in the coming weeks. Time will tell

16 April 2013

Logs indicate that in the last two weeks I used rsync to pull stuff onto my phone three times as often as I made voice calls. Basically busybox/rsync/sshd are the holy trinity of catnip and if my phone's dialer broke I probably wouldn't notice. This is a good thing.

12 April 2013

I recently became an Amazon Prime user, getting a free ride from @IBMaxP. So naturally I've been ordering more from Amazon lately. Yesterday I was ordering a bluetooth adapter for my new raspberry pi (will I stop at 2? I doubt it), and they were £1 on ebay and £2 on Amazon. That's close enough for Amazon's superior logistics and delivery times to win them the order.

Fast forward 12 hours and I have the adapter in my hands. I also have a large box and a lot of padding, all of it having been delivered by courier this morning. I know that the Prime service relies upon charges being averaged across a large number of users but this is insane. The adapter would easily fit into a padded envelope, and could probably have made it here "next day" via the postal service.

And the actual bluetooth doodaa? Dead on arrival. I should have stayed in bed.

02 April 2013

The more old photos I look at the more I want to spend my weekend's trawling junk sales for old crap. The Fiske Reading Machine, held here by inventor Admiral Bradley Fiske, looks like the kind of thing I need to own. Not "use", that'd be ridiculous, just "own".

28 March 2013

A few days ago I mused that I may be the slowest bitcoin miner in the world. This is exactly the kind of piffle that I come out with all the time, and invariably I am wrong. I am not the slowest. I am not even in the same ball park.

Mining on a NES. That could be the slowest, it really could.

28 March 2013

Way back on the 21st Nov 2012 I backed a Kickstarter project for a little machined metal key-fob case. The case arrived this week, and as there have been some very feisty comments on the project's page I thought I'd post my thoughts.

Lets start with the bad - the o-ring is very poor quality and over-compresses / shreds with even light tightening. I won't be using it. The supplied key-ring is also a little too large for the mounting hole, so I'll be replacing that too. Both these failures are pretty disappointing, because they're the result of a lack of care on the part of the maker. Its clear that he didn't trial multiple choices and then select the best option. Instead he sourced some parts (probably based on price) and then shipped them even when they weren't quite right. But... lets not forget that the o-ring and the keyring are by far the least important part of the product.

Next the body of the case. The vocal complaints coming from other buyers are that the case isn't correctly machined and isn't correctly polished. I can see the basis for these complaints, but I have to say the quality is "good enough" for me. My case hasn't been finished to a high standard, but its pretty good. I had some black grease on the threads and the inside is pretty rough, but the outside of the main body is actually fairly nice.

The material is supposed to be titanium and I'm happy to assume this is the case. I have some other titanium items and the colour is a good match, and the weight feels about right in the hand. I haven't taken any measurements or tested hardness, it just looks about right. You can see the banding from the machining process, the surface hasn't been polished well (if at all), but there are no major dings or scratches to complain about. I didn't have any plans relating to the exact dimensions or internal bore, so can't comment of whether its ended up being to spec or not. All I can say is its a perfect match for the height of my car key and it holds what I want it to hold - so by chance or design the size is correct for me.

As mentioned above - the inside is rough and greasy. My flash drive fits nicely once I clipped a little of the plastic off, so I'm happy. I don't need waterproofing as the car key it'll share a fob with isn't water proof anyway. The treads match nicely and screwing / unscrewing the cap feels solid and assured. Once its shut it grips well and doesn't work loose.

Overall I have a product that I'm happy to use. It'll be on my keys until I get bored of it or it breaks.

So do I feel cheated? Scammed? No I don't. There are certainly people over at Kickstarter that feel that way, but I don't. This is just a flawed product, where the execution wasn't as good as the idea. Its "good enough" and many people will be fine with it. There seems to be variable quality (perhaps mine is a good one) and poor communication - which never helps. I just feel that with Kickstarter you're making a judgement call about the project creator as much as anything else. If the driving force behind this project had been an artisan who cared deeply about the product then these items wouldn't have shipped. There'd have been more preparation, better components and better quality control. This is a simple product after all, a skilled and caring creator could easily have got this right. But just because the project fell short of that it isn't necessarily a scam. Not everyone has the talent to master the manufacturing process.

My key-fob is of reasonable quality and was moderately overpriced. I'm happy with that.

25 March 2013

BS 1363 is the world's pre-eminent power plug standard. Just compare it to the pitiful scrawny things our American cousins have to put up with and you'll be singing the national anthem and making an impromptu speech on the merits of empire. It may be bulky, and the most painful item in the world to step on barefooted, but it's safe - Grounded, fused and safe.

Except it isn't.

I honestly think that cheap plugs are being shipped fitted with placebo fuses. They never seem to blow when some errant equipment goes berserk, the breaker in the distribution board is always the one that pops. And then we have delights like the one pictured above... rated for 10A, fitted with a 10A fuse and it was happily running an 8A dishwasher despite having melted into the socket. I think the heating coil in the dishwasher has failed in a dangerous way, apparently they're common causes of house fires. The plug smelt terrible, maybe the smell is intended as some kind of early warning signal, which goes off a few minutes before the fuse pops...

22 March 2013

White balance, I don't really care.

The accepted wisdom of digital photography is that white balance matters. When you take a photo the camera makes a guess about the colour of the light illuminating the scene, and then corrects the exposure to ensure that the colours of the scene are correctly recorded. So if you use your camera's flash (which has a blue tint) to photograph a sheet of white paper it should look roughly white on the resultant picture. And if you lit the paper with a tungsten bulb instead (fairly yellow light) then it'd have to correct for that to give you white paper in the image. This process doesn't just apply to the whites in the scene, they're just the most obvious, it is actually applied across the entire image.

The problem with this process is that it doesn't work. Modern cameras are a lot better than earlier ones, but they're still really bad. Even if you prod it and supply a hint by manually selecting the light type (e.g. sunshine, or fluorescent lights, or flash) it still gets it wrong as often as right. The correct way to deal with this is to shoot in RAW mode and apply colour correction manually in post-processing. 95% of the good looking images you see in printed media will have been balanced manually.

But I hate white balancing, it bores me and my tools (basically Gimp) are pretty poor at it. Worse still - I don't actually like the pictures more when they're correctly balanced than when they've got wacky levels. Well, most of the time. A particularly good picture will normally benefit from thoughtful colour correction, but for random quick snaps like the ones I post here I prefer overblown highlights and hopelessly over-the-top saturation. If I post 3 photos all against the same backdrop you'll probably end up with 3 different colours. And lots of lost detail because the contrast has been overdone.

Ramble ramble. All I'm trying to say is that I know the pictures on this site are all over the place technically. Its a deliberate decision, even if it looks like an error. I don't have the time/tools/ability to do it correctly, so I do it oddly. This is my modus operandi in many areas of life, and I'm comfortable with it.

21 March 2013

My daughter Mia is very interested in science. I am very interested in things that can be made to go boom. So its pretty obvious that we need to make a device that can convert water into H2 and O2, i.e. electrolysis of water.

In order to produce the gases all you have to do is arrange for the water to conduct electricity, stick a pair of electrodes into it and pass a current between them. The gases will collect at the electrodes. If you want to separate the gases you con enclose each electrode in a collection mechanism, if you just want a mix of gases then you can just collect the whole lot together. I want the mix, because collecting it this way ensures its in the right ratio for those exciting booms.

In order to make water conduct electricity easily you need to create a solution of charged ions, by adding an electrolyte. I decided to use potassium hydroxide. The first thing to point out is that KOH is fairly dangerous. If you get it on you your skin (or worse, your eyes) then you're going to really regret it. Safety first and all that. But the upside is that it behaves well as an electrolyte and you get clean products. The naive way to do this experiment is to use everyday cooking salt as the electrolyte, but this will yield chlorine in your gas products. By using KOH we get a fairly efficient reaction and quality products. When you dissolve KOH in H2O you get K+ and OH- ions, and a solution of ions will conduct electricity.

KOH(s) → K+(aq) + OH-(aq)

This process is exothermic, add the KOH slowly to avoid heat build-up. KOH is, of course, cheaply available on eBay.

Researching water electrolysis on-line is a rabbit-hole of crazy. Large numbers of crackpots of every description seem keen on the process. Mostly they think it'll solve all their gasoline dependence problems. Build one of these devices and strap it to your V8 truck and you'll be using less petroleum than a granny on a scooter. It'll be useful when the government breaks down and we all have to retreat to our bunkers. Well wingnuts they may be - but they have some decent engineering ideas. The first idea I'm going to steal is to replace graphite electrodes (as used in the high school chemistry version of this build) with steel kitchen supplies. The main reason to do this is cost. If you get good quality steel plates they're also very reliable, but I'm using incredibly thin steel of unknown composition... so cost can be my only justification. These cheese graters were a few pence each. Bargain.

Eventually I'll have 3 pairs of cheesey electrodes in a row. This'll give a nice high surface area and a low separation distance. For this prototype I'm just using 1 pair.

It was at this point that I was able to properly explain the process to Mia, with all the constituent parts were at hand. We talked it through and played with the ideas of atoms, molecules and chemical processes. I suggested that Lego could be used to hold the plates and she made this marvellous frame. I attached some supply wires to the plates and then submerged them in the solution.

At this point we would have been delighted to see massive effervescence as the house rapidly filled with a stoichiometric mix explosive gases. But that didn't happen. The goal of this first session was to explain the idea of splitting water into its component parts using electricity, not to actually achieve it. I hadn't sourced a power supply for one thing so we were limited to using an old 9V PP3 battery of unknown provenance from a kitchen draw. We saw a really small amount of gas being produced, enough to claim success, but nothing worth collecting and playing with.

We'll be returning to this project next week when Mia has some time off school and Rebecca won't be around to complain about our danger-lust. One thing that came out of our discussion was Mia's interest in seeing if we saw twice as much Hydrogen as Oxygen, based on the water molecule's components. This was her own prediction and a great insight, but unfortunately we won't be able to look for it as my design mixes the gases up. She was so keen on the idea that I may get some alternate electrodes so we can try it. I also need to arrange to have some Jean Michel Jarre playing next time. There is no point in doing these things half heartedly.

20 March 2013

I may have mentioned it before - pencils are ace. For the last year or so I've been mainly using a KUM Stenographer's Long Point Sharpener to keep mine honed and ready for action. The Long Point Sharpener has the advantage of giving you a really drawn out taper to the point, which I find very satisfying to work with.

Recently I decided to give the slightly more involved KUM Automatic Long Point a go. And its excellent. Really quite dangerously pleasant to use. I've been sharpening my point between words, this thing is so nice its borderline counter-productive.

The operation of the sharpener is slightly different from a normal one. First you sharpen in the left-hand hole, which removes wood from the pencil body but leaves the graphite core untouched. As you do this a long (but blunt) nib forms, and eventually runs up against a wall. Once this has happened the pencil will just spin without removing any more wood. You then move to the right-hand hole, which only sharpens the core and leaves the body untouched. A few turns puts a nice point onto the pencil and then it will just spin. You can't over sharpen, which is my typical error with the traditional design and gives you lots of broken points.

The mechanism is supremely tactile, a quality piece of kit despite its plastic body. There is, however, a drawback. If I'm entirely honest the results are a little too sharp, its actually quite difficult to get them started on the page without snapping them. But if you like your pencil to look like a weapon as it sits in your stationery pot (and I certainly do) then this is the tool for you.

17 March 2013

I have discovered an amazing new way to end up with a crap product when buying on ebay - when the product is actually as described.

I buy lots of crap from ebay. This makes me very sceptical of sellers' claims, because a lot of them just lie. So if an electronic gizmo is very cheap and proudly claims to be posted from within the UK you can bet it'll be posted from Hong Kong, and will take weeks to arrive. No problem if its cheap, just factor the time into your choice. And if they claim the item is made from spacecraft grade titanium (whatever that may be) then it'll actually be made from cheap plastic.

So imagine my surprise when the phone case I bought turned up the next day, and really had been posted in the UK. Even more shocking (given the cheapness of the item) - it was made from what looks a lot like the "brushed metal aluminium" they claim. This is, frankly, a disaster because it blocks the WIFI signal almost completely. Cheap plastic would have been far better. The case looks fairly good, in the loud and brash way I like, and it feels great in the hand... but I wanted a case not a Faraday cage. So it goes.

16 March 2013

For the last 3 weeks I have been mining bitcoins on my raspberry pi. I suspect that I may be the slowest miner currently active worldwide. I have no proof to back that up, but it amuses me to thing it might be true.

A long time ago it just about made sense to mine on a CPU. The rewards were fairly high and general purpose CPUs weren't that much slower than other execution unit choices, not once price was factored in. It didn't take long of GPUs to take over as the only sensible choice, they have a very large number of pipelines and are very efficient at running the hashing algorithms. A few dedicated miners have made good use of FPGAs but GPUs have been the sweet spot of convenience / price / performance for a while. We're now on the brink of the next shift, away from GPUs and into ASICs. Purpose designed bitcoin mining ASICs are launching right now and they're a lot faster and far more efficient in terms of power (one of the biggests costs to consider is power) than anything else on the market. Once they're widely in use people mining with GPUs will be fighting a lost cause. Anyone still using CPUs is already in this position. Anyone running on a slow and inefficient CPU is doubly so.

Which brings me to my mining rig. 187 kilohashes a second of pure glory. Miners on x86 platforms have the benefit of decent algorithm optimizations such as sse2_64, we ARM folk have raw C.

My desktop machine, running with sensible priorities so I can still use the machine, clocks in at 47 megahashes/sec, 250 times faster. I've seen an ASIC user clock 4,800 Gigahashes/s - 25 million times as fast. So yeah, the pi is slow. How slow? Well plugging in the current difficulty (4,847,647) and my speed yields an "expected" time of 3,530 years, 201 days (can vary greatly depending on luck) per block generation. That difficulty value will probably jump up once the ASIC players are online, so this will only get worse. Thankfully the use of mining pools means I don't actually have to generate a block in order to see a return. For the fee of a few percent I can get paid on a pro rata basis for my work towards a block's generation. I've joined the BTC Guild and expect a fat payout any day now.

15 March 2013

I was vaguely aware of the kickstarter campaign that Anita Sarkeesian started to raise funding for her video series examining the roles of female characters in gaming, but I didn't actually back the project. I hadn't seen any of her videos and hadn't really thought too much about the subject, at least not in more than a superficial manner. Fast forward the best part of a year and she has taken the ~$160k raised[1] (against a $6k target!) and put it to good use. The first of her videos is here: Damsel in Distress: Part 1 - Tropes vs Women in Video Games. I thought it was great. Really worth the time to watch, and I'll be sure to catch the others when they come out.

1: Her target of $6k was sensible and would probably have been exceeded, but the torrent of misogynist hate that rained down on her and the project made sure of it.

14 March 2013

For several years now people have been saying that Google Reader was on borrowed time and was overdue being turned off. I ignored them. Now we have the official confirmation that the plug is being pulled (on July 1st 2013) and I've gone into some wierd kind of regretful shock. The main feeling is that I should have knuckled down and written my own replacement years ago when I had the chance. This is my standard response to most things so can be dismissed. The other feeling is anger, which isn't something I've drawn from any previous closures I can recall.

Normally I'm fairly pragmatic and just accept the changes with a laissez-faire air. It wasn't my service, it was free, it wasn't very good, what does it matter? Well this time it does matter. My absurd levels of content consumption are only possible because of an efficient work-flow, and Google Reader was the absolute keystone of that process. I'm an RSS junkie, and I'm not sure I could survive withdrawal.

A few minutes with a scrap of paper is enough to convince any sane thinker that a decent RSS feed reader is actually a tricky engineering problem. A toy one can be cobbled together in minutes, but anything approaching the power of Google's implementation would require real resources behind it. So rolling my own isn't something that'll happen tonight. I'm going freemium instead and have thrown some money at Newsblur. Good luck to them. And they'd better not let me down, another disappointment on this scale might be too much for me to take.

13 March 2013

I quite fancy one of these 1936 50th anniversary coke calendars, but they appear to sell for several hundred dollars in good condition. I don't want one that much. Which leaves me slightly jealous of this dude lording it over me with three of them.

03 March 2013

I am somewhat renowned for suffering from not-invented-here syndrome. By that I mean that I dislike using software tool-kits without at least a degree of re-inventing them. The strongest manifestation of this has been blogging platforms. I opened a Blogger account sometime late in 1999 and instantly took a dislike to it. Instead I wrote a horrendous content management system in Modula-2 (a wholey unsuitable language for the task) that spat out HTML files which were then sent to a webserver by FTP. Since then I've authored roughly as many blog engines as I have blog posts. They've all be pretty flawed, usually built around some quirky core design choice that I selected because I'd never seen anyone else do it. This isn't always (or even often) a good place to start a build from.

My first serious blog platform was probably my 5 or 6th, I think it launched in about 2003, The colablog engine. It had a decent template engine and ran about 20 domains, about 30k posts and perhaps 2 million page views on a good month. More recently The wiaf engine (I have a talent for snappy names) is still live and currently hosting over 20,000 active users with approx 4 million page-views a month between them. Its a CMS really, but it started out with lots of blog inspired ideas that slowly got swamped by other stuff.

But I never write any content. I add new features, but never new content. So this page is becoming an attempt to push back against that. No blog engine. No RSS feed. Just a single HTML page with a load of old crud in it. Like Geocities. Hurrah.