Speaking to a friend who's a trance-techno DJ, he's hurting because vinyl is getting so expensive. 10 Euro, he says, for a 5 minute piece. But a working DJ needs dozens of these each week, always the latest and newest stuff.

Pioneer have a cd-based turntable that DJs like, but it's not the same. Good, yes, but it's not as good as vinyl.

So, this prompted MDPA/4. The concept is a modded dual-platter turntable, consisting of:

  - laser diode and reader mounted on the needle heads
  - rotation sensor on the playback arms
  - internals augmented with a digital music player
  - small touch screen allows selection of left and right tracks

The original audio circuits are completely disconnected, although in principle they could be left in place for the fun. The laser diode and rotation sensor tracks the movement of the head across the vinyl and this movement (track play, skip left/right, back, slow/fast) is translated by the media player into a realistic simulated sound.

If the DJ picks up the needle and drops it again a few tracks further, the digital playback follows. If the DJ scratches the music, the playback scratches. If the DJ slows or speeds up the record,... you get the picture.

The advantages of this design? The DJ gets a pure vinyl feeling, finger-tip control over that glorious black plastic. The record companies and artists producing the original pieces have a much cheaper distribution mechanism.

Implementation of the media player: probably needs to be based on a lossless compression format (even WAV files), and using a harddrive for storage. The use of a lossless format may help with the main problem, which is to ensure a perfect playback with no noticeable lag.

The player would have to be calibrated for a particular vinyl. This is done simply by playing the record once, so that the needle and rotation sensor can track and map the surface. This mapping can be overlaid onto the track being played so that a perfect match can be made between the position of the playback head, and the portion of music being played.

Optional extras: a CD reader which loads tracks of a CD, digitises them, and adds them to the database for later selection.

Price: should be doable for about $200-400, depending on the amount of storage provided.

Instant poll:

Who smoked the most crack in 2003?

    (_) SCO
    (_) Belkin
    (_) Verisign
    (_) *A
    (_) All of the above
    (_) CowboyNeal
    (_) YIS I can't afford crack!

This one comes from a discussion on /. today about spam.

Spam, spam, glorious spam. It's not getting any better, despite creative solutions of all kinds. And how can it? For every inventive and courageous spamfighter, there is an equally desperate and inventive spammer. Necessity breeds invention and what man makes, man can hack.

To me the end game is clear: all data transferring in through the firewall - email, HTTP, ICQ, ... - will eventually be treated as hostile unless it is assured to be safe.

The days when the Internet was a global village happily exchanging high moral values by NNTP are dead and gone.

The trick will be to extract the legitimate data from the mass of corrupted garbage that will wash around the fibre optic oceans of the Net. Data mining? I already do this... over a thousand spams a week, and somewhere in there are several dozen vital business and personal emails. I miss a few each month, it's rarely tragic, but this is despite having the latest tools (SpamAssassin) to help.

A consumer PC is infected before it can download the Windows patches it needs to be "secure". 90% (99%?) of all new home PCs bought this year will be infected before their owners have time to click 'Windows Update'.

My solution is a system of data delivery via a global trust network. Six steps around the world. If I want to send data to someone, I send it to my data broker. This is a company that spends its whole time checking its client list: clients are vetted, must pay deposits, and generally treated with the same paranoia that a bank would treat a new client asking for a home loan.

You can't simply connect and start sending data. When your reputation is low, the price is high. As you build up a good karma, your price drops until it's very low or free.

The broker, in turn, speaks to other brokers and passes your data along in turn. Brokers pay for this, so your money is actually used to finance the trust network.

You can choose brokers: the most highly rated are also the most expensive, but you will be certain that your data will arrive, because no-one will refuse data from an AAA broker.

The system does not require micropayments or any other complex financials: you either pay upfront for a bulk traffic package, or pay an invoice as you would a credit card. As an unknown and untrusted client, you will have a tiny allocation that you can abuse with little risk to the broker. As your reputation (with that broker, mind you) improves, you get larger and larger allocations.

It's up to the broker to implement the necessary checks and balances so that a spammer cannot create an account and then cause havoc.

Now implementation. This has to happen at the packet level, cover all protocols. All data has to be encrypted between parties, so that it's impossible to someone else's identity and steal their reputation. And it has to happen in realtime, without slowing down interactive use of the web.

I think the parties best placed to make this happen are large web email providers, who can implement 'guaranteed email' delivery to and from their clients. This will be a paid service, spam free, and with delivery reciepts.

That's the idea for now.

Cars should be able to show emotions. I'm not just talking about the brake lights that shine brighter when you brake harder, I'm talking about a fully controllable paint job that can flick from blue to red, to green and back.

When some jerk drives too close behind me, my car will start to glow angry red. "Back off, dude!" When the cops are prowling, the car goes discreet black, "nothing to see here, move on."

When it rains, the car will turn violent neon yellow, visible from a mile away. When it's beautiful blue weather, it will turn a reflective silver, a gleaming mirror.

And finally, when it's parked in my street, in one of the worse areas of town, it will look rusty and cracked, a poor man's car.

Implementation: an all-over digital skin made from the same stuff those building-size disposable flatscreens are made of. (Yeah, we're 2010 already!)

This will be an irregular section for wacky ideas that I want to publish before some jerk places a patent on it and steals my millions. There's a logic in there somewhere... Anyhow!

Today's entry: self-rotating web adverts. Yes, you read about it here first! (PatPend Heironymous Coward 2003).

Description: instead of showing a new banner ad for each new web page, the advert shall be reloaded au-to-ma-ti-cal-ly every ten seconds using (gasp!) a JavaScript timer (gosh!). Yes, the logic will be to wait ten seconds, the fetch a new random advert from the ad server.

Advantages: many more impressions per viewer, and direct targetting for the morbid crowd who have nothing better to do on a cold winter evening than to watch the banner ads. "It's Art, Jim, but not as we know it!"

Implementation:

var the_timeout = setTimeout ("javascript:refresh()", 10000);

That's all from More Damn Prior Art for today. Tune in again soon for more news!!!

Heironymous' Thought for the Day:

"You can measure a man by the size of his lawyer."

Context: Heironymous is learning about company law the 'interesting' way, and particularly enjoying the ongoing chessgame between the two lawyers. Mine is going to win, I can see it. The other has already regretted starting the game.

'Nuf said about this subject.

On Sept.12 2003 I discussed a new kind of user interface that I wanted to build. Well, we have started on a prototype and it's quite nice. My goal with documenting this as we go along is to get feedback (obviously) but also to make very certain that there is prior art in case someone decides to patent the concepts.

I'll just have to describe it verbally for now, screenshots and a test version will be along when we're ready.

The UI is the simulation of a real workspace, a desk top, and I supposed you could say its logo is "A tidy desk is a sign of a sick mind."

The concept, which we're calling "Clutter" for want of a better word, is simple and yet general. First, the desktop is just a large space with random icons, which we call "motes". A mote is represents a link to some resource: a document, a directory, a web site, a BitTorrent link, whatever. Anything you can drag and drop can become a mote.

Initially, motes look like normal icons on your desktop: they have an icon, a small text label, and they position themselves so that they don't overlap. If you drop a new mote down onto a space filled with existing motes, they will rearrange themselves to make space.

Here is the first innovation: as you add new motes, older unused motes shrink in size. This happens progressively so that as motes go from 'hot' to 'warm' and then 'cool' and 'cold', they shrink from full-sized icons to half-sized icons, and finally, simply little dots. The text label disappears when the mote is no longer 'hot'.

Motes use color to signal different states. This will become more useful and necessary as the intelligence of each mote increases.

As motes shrink, they also demand less space around themselves. So a bunch of motes that are ignored for a while will collapse into a dense cluster of dots. This happens over days and weeks, and the desktop ends-up looking more like a simulation of some articifial lifeform than an organization of a desktop.

You might think that a bunch of unlabelled motes is just clutter, and perhaps it is, but we've found that it's easy to remember what things are by their relative position on the screen.

The mouse acts as a kind of magic wand. Waving it over motes causes the motes to highlight, and the text label to appear. After a small pause, which decreases as the desktop gets fuller, a preview (of the document, web page, whatever) pops up. A click opens the document for viewing, a double-click activates the mote and brings it back to 'hot' status.

Motes can themselves be desktops. It's easy to create a new embedded desktop by collecting a bunch of motes. By embedding desktops recursively, we get an infinite space to play with, and an easy zoom-in, zoom-out capability that can be used to simulate a classic hierarchical filing structure, multiple desktops, etc.

Desktops are defined as a large set of XML files that can be manipulated by external programs. Since motes can be executable programs, this is convenient. For instance, I am making a mote program (perhaps the word 'agent' is appropriate) that checks my email, scans against a whitelist, and drops all priority messages onto my email desktop.

There is a search function that highlights all motes that match. Simply and instantaneous.

The second innovation is this: motes can be scripted to do useful things. This aspect still has to be designed and tried, but my proposal is as follows. Motes are implemented as XML objects. All motes have data that is standard and used by Clutter. Then, motes can have their own data, which is application dependent. An email mote, for instance, has address information, list of attachments, and so on.

Motes are always based on a mote class, itself implemented as an XML file. The mote class provides the scripting, which is based around a workflow model that I will explain when we get there. The goal here is to make it possible to build the logic for (e.g.) email handling into the mote class, and have this interpreted by the desktop.

We have more ideas for v2, but this is already a lot to be starting with. Clutter will evolve slowly as we use it, the ultimate goal is to use it as the basis for a new kind of workflow-based application interface in which a single screen unifies everything you are working on, from personal projects and family photo albums to web sites, incoming emails, requests from clients, etc.

Comments welcome.

I used to write a webzine, and one of the fun parts was to make a set of predictions near the end of the year. Usually I was right, only a year or two too early.

Well, here is my list for 2004 and onwards. I'll make my bold assertions first, and explain them afterwards.

1. Linux will wipe Windows off the commodity IT landscape. Like the tabacco firms, Microsoft will find itself assailed from all sides until it retrenches into a few core markets (in this case the US), and either falls apart or moves to new turf.

2. The most exciting firm to watch in 2004 will be IBM.

3. The killer mobile application of 2004 will be simple SMS. MMS, GPRS, WAP,... not important.

4. The most interesting market (both for consumption and production) in 2004 will be China.

5. There will be no significant financial recovery in 2004, but no further crash either.

6. The software world will discover code generation in 2004.

7. 2004 will herald a new technology revolution based around the battle between those who seek to control the price of technology, and those who seek to mass produce it. FOSS is just one of the symptoms of this battle.

OK, now my reasoning.

The most important driver in business is the cost curve that mass production creates. This is well known in IT as Moore's Law, but I believe it applies as much to software as it does to chips. The cost of producing software is a function of the quality and availability of pre-packaged pieces. I see this as a smooth curve where prices halve each year, just as for chips.

Inevitably this means that the software barons are already dead. It is difficult to sell software when your competitors make it better at half the price. It is impossible when your competitors make it ten times better at one hundredth of the price. We are not yet there, but we are moving that way.

Microsoft find itself on the wrong side of this curve: every improvement in software packaging and distribution - and I'm thinking of things like Debian packages - is a nail in the coffin of Microsoft's business model. Cheap, reliable, simple, powerful software is a terrible thing to fight against. We were surprised when Munich chose a FOSS strategy over Windows. But the dripping has turned into a steady stream, and soon it will be a waterfall, as institutions and governments around the world ask themselves three simple questions:
1.Do we really need the latest and greatest in software?
2.Can we save money by taking something simpler?
3.Will we be taking any serious risks?
And the answers are of course: no, yes, no. Enough for any decision maker. Add a little WTO pressure to reduce the use of pirated software, and the balance tips.

Anyone who still thinks Linux/FOSS can't compete against Windows has a shock coming. In my small company, I find myself responsible for installing machines. What used to be a complex business of finding the right device drives, the correct versions of applications, often in various languages, has now become an exercise in simplicity. Pop in the Xandros CD, install, download OOo, install, configure email, and give to user. Finito. If this works for me, a professional technologist, it most surely works for a billion late adopters around the world.

Microsoft's biggest weakness, and this is so shockingly obvious you wonder why they have not addressed it, is that they make products for the early adopters. Each release of Windows and Office sells on technological merit, not on stability or cost. And yet mass market consumers - which means everyone who does not care what's in the box, so long as it does what it should - buys on the basis of cost and economy.

Microsoft should have a series called "Windows Classic", "Office Classic", which sell for $49 and which provide 1999-vintage functionality in a smooth package that installs and runs on any hardware with no questions asked.

I still can't believe that plugging a new mouse into my Windows laptop results in five different dialog boxes, while plugging the same mouse into a Linux box results in exactly nothing except a half-second delay before it all works perfectly. And this is true for USB dongles, new network cards, DVD drives, etc. etc. etc.

Joe Normal, trying to start a small business in Ho Chi Min City, wants a PC that does exactly four things: email, web, office, and possibly chat. That is the global commodity IT market, and it's one that Microsoft does not "get".

IBM, on the other hand, gets this so well it's scary. They don't even need to produce an own-label Linux - though they do and will probably provide one for consumers eventually.

Now onto mobile information. I've been looking at the market in Europe, one of the more mature ones. It's no coincidence, but large suppliers still try to sell new products based on technological superiority. You may ask, "what other basis is there"? The answer is, "sell what people want." My firm makes SMS applications. Not simple ones, but fully interactive SMS gateways. Want to book a ticket via SMS? Want to find someone to share travel costs? Perhaps check what movies are on this evening? The more we look at this market the more we're amazed by it. Something like 70% of people in Belgium have GSMs and regularly send SMS messages. But they are offered almost no applications at all beyond logos, ringtones, and chat. My younger sister has no email address: she does everything by GSM. For many people today, their handset is their terminal, their only connection to the world of information. For these people, SMS is simple, obvious, and omnipresent.

In their rush to push growth, the operators have ignored this, and instead focus on new gadgets: video phones, portable cameras, 'picture chat', and so on. It's fun, and it will drive some GSM sales this Xmas, but it won't touch more than 2-3% of the user base. The concept of using SMS to recreate a kind of simple mobile internet is both so simple and so useful that I predict it will become a hit business in 2004: where people used to flash their web site address on adverts and business cards, we will see people flash their SMS site address instead. My company is just launching such a product - www.sms-at.com, you need a Belgian mobile phone number to play.

Now China. All I can say is look at the trade figures, the amount of exports, but more importantly, the amount of imports. China is no longer a virtual market, something that may happen one day. Today it already imports more than it exports, and if this is not a lesson in the benefits of trade, nothing is.

When two countries trade, both benefit, though jobs are often lost in the process. When the entire globe trades, there are real and long-term benefits, and I believe the benefits of the last ten years' globalisation will become clear as we see that ever cheaper goods and services make life ever cheaper in Europe and the US even if wages don't rise.

Code generation is one of my pet magics. True magic. Tools like GSL, which I've discussed before, turn the business of software development around from one of artistic creation to one of industrial production. I believe this is one of the most revolutionary technologies in software development. The increasing number of books on the subject, the increasing number of people who actively use this technique in their work, and discuss it in blogs and conferences... it looks like an awakening is in the air and with a little luck and hype, next year will see code generation move into the general awareness.

Lastly, the digital divide, the battle between those who think culture is something that can be owned, and those who believe culture has its own existence and that we're just its temporary guardians. Both sides are right of course, this is the nice thing about truths, there are so many to choose from. But the hoarders have been outfoxed by technology, the internet and the hundred-gigabyte hard disk. Game over, check mate, however you turn the egg around, it has been solidly cooked.

The nice thing about human nature is that it works all by itself.

Found this interesting take on the IT industry.

And I'm still wondering how I can find back that gem I wrote around comment nr.24. Even Google can't find it, so it's really lost.

Deep somewhere in the dark depths of the Slashdot database there sit my 400 comments. All my wisdom, poured straight onto the web, and now lost forever.

There's a serious point trying to make its way through this shambolic excuse for prose. I believe I'm one of the great thinkers of our time, a true 21st Century Renaissance Man, with an opinion on everything and occasionally something sensible to say as well. Posterity will treasure my words... well, they would, if they could find them.

Will I be forced to bookmark my own comments so I can find them again later? Or perhaps cut and paste the ones I find tremendously witty and insightful (usually those that get moderated straight down to -1 Troll)?

Let me make a small diversion before coming brutally to the point.

There was once an ancestral human population that appears to have lived along the coast of the Indian Ocean, stretching from Arabia to Australia, adapted culturally and physically for the hard life of seagoing fishermen. Between fifty and ten thousand years ago, this culture refined the technology of navigation that let them colonise as far away as the pacific islands, possibly even South America across the Pacific ocean.

We know they existed because our genes show the traces, especially in the sea-burnt black skin and tightly curled hair that is still common all around this ocean basin. We assume for some reason - climate change, possibly, boredom with seafoods perhaps - they dropped their costal lifestyles and emmigrated inwards, settling Australia, Indonesia, and above all, Africa, pushing the existing Khoi'San inhabitants inwards and westwards until there remained only pockets in the jungles and deserts, the pygmies of the equatorial forests and the San of the Kalahari.

Few traces remain of this culture: only a few tantalising hints, such as genetic protection against malaria, some common word roots, some sailing and navigation technology, and some musical instruments like the thumb piano show the common origin of cultures as wide and varied as the Bantu in Africa, the Dayaks of Indonesia, and the Aborigines in Australia.

And this entire culture came and went after Europe was settled by its present people 100,000 years ago. Africans, instead of being the original inhabitants of Eden are its most recent colonists, possibly the most striking and modern adapted ethnic group.

Or perhaps the whole thing is a fantasy. We'll never know, because, like the history of the 21st century, their story was written in beach sand, and washed away by the next tide.

There is a Google search that lets me find my old articles. From 400, I find exactly 45.

It was a sad story today about the lawsuit against Sony following the GTA murder case.

But also sad to see the many "problem with parenting" comments that followed. So, read my comment on what I believe actually lies behind the murders, and behind many causes of youth violence (Columbine High included).

I've always found that blaming others for problems is just a way of avoiding any action that could cure the problem.

So, read my comment and then tell me what you think of my programme for slowing down youth violence.

Step 1: determined decriminalization of drugs and any other behaviour that lies in the "individual choice" category. Purpose: to reduce the criminalization effect that puts many young people outside the law.

Step 2: create a branch of the justice system (police, prosecution, and courts) specialized in tackling crime by young people against young people. Penalties should be clear but always based on forms of social redressment. The mentality of "leave me along, I'll handle it myself" must be softened and finally eradicated.

Step 3: create social structures that bring young people into contact with different age groups: for example, tie education into professional training and shared learning so that young people learning some skill can spend time both with older kids who are perhaps already working, and with younger kids, who need coaching. It could be music, sports, auto repair or any of a host of other skills.

Step 4: use genetic screening to determing kids that are "high risk". There are specific genetic markers for violence, the simplest is to monitor adult males in the same family. Young boys who are at risk in this way need special attention and possibly alternatives to a normal civil life. Such men tend to make excellent professional soldiers, for instance.

Step 5: understand that the education system is broken in many ways, and fix it. First and most importantly. align school hours with professional life so that kids do not return at 3pm to an empty house. Secondly, abolish the long summer holidays and allow kids to take (e.g.) 20 days' holiday at any time in the year. Thirdly, introduce vertical education groups so that children spend time with older and younger children (as I already mentioned). There are a host of other problems with mass education, but these are the main ones.

Lastly: repair the enormous adult-youth gap that exists in the States today. Find ways to create a single human culture, based on local communities, strengthed by a common sense of purpose and responsibility, and bolstered by laws that protect but do not alienate.

As part of my ongoing (interminably so) study of human nature, I am defining an index that I call the "Internal Mortality Rate", or IMR.

The IMR of a group is that percentage of its members that are imprisoned or die in any given year. Strictly speaking there are two IMRs, one for incarceration (call it IMR/i) and one for judicial killings (call it IMR/k).

The IMR of a state is a direct and objective measurement of its degree of repressiveness. One should be able to plot a state's IMR over time to understand whether it is moving towards or away from improved civil liberties.

Increased IMR/i is a bad sign, it means either that large segments of the population are being criminalized and locked-up, or that the state is on the way down the "death-squad" road. It's possible to analyse this by looking at the profiles of those being locked up. If they mainly belong on one social or ethnic group, we have an instance of social disharmony that can lead, in extremis, to genocide. If the victims are mainly the young educated urban elite, we have an instance of the old vs. young generational power struggle, typified by a socialist vs. facist struggle for power. In which case an increased IMR/k is highly likely.

Most states on an ascending IMR curve will eventually limit press freedoms so that the public awareness of the actual situation is controlled. However, this act is generally delayed, giving the serious social scientist a window in which to measure the IMR.

I propose, therefore, a general study to rank the 152 countries of the UN according to published and estimated figures. This should be possible, and would be valuable information for human rights activists seeking to learn where to focus their energies.

There is only one serious problem with Xandros, and I'll get to that.

I'm not a Linux guru, despite having used it for a long time. I've installed perhaps twenty PCs with various distributions, starting with a Yggdrasil CDROM sometime in the 90's. It was a solid installation that did exactly what I needed: act as a server on which I could compile and test my free software tools. Even then, the fact that I could load a PC with free software capable of acting as a full development workstation for my C code... it seems banal now, but those were the days when compilers were big business.

Over the years I've tended to delegate Linux installs to younger, nimbler minds. I've admired how they played with disk tables, network configurations, not to mention loadable modules, and darker magics which my mind was quick to erase.

Then, last week, I found myself faced with four PCs to install, two for family members, two for new colleagues. A couple of decrepit workstations, which we bought in 2000, barely able to boot without pausing for breath. And a couple of notebooks, Austrian boxes built from desktop components, so cheap and nasty that out of an original four, only two survived more than a year, and on one of those the display will only work after a good hour of warming up.

My users have simple needs: browsing, office. Linux, yes, with OpenOffice.org, which we all use here. My mind waters at the prospect of desktops with no malware. No trojans, no Kazaa, no porn dialers... YES! We have been eliminating our Windows desktops, but each new install has been an adventure in itself.

Of course, I wanted to install Debian, there is no other Linux than Debian, as my sysadmin has told me many times. "Don't even think it..." Seeing how he can manage several dozen servers with just the lightest of touches, I have to agree.

Debian... now, I'd bought Lindows, which is based on Debian, a year or so ago, and it is really a cute package. People will mock it, but Lindows does so many things right that it's a kind of jealous mocking, "hey, why can't I download it for free, huh?" Lindows installs incredibly quickly and efficiently. None of that sick "and now you will configure your X display" nonsense.

But I'd already tried to install Lindows on one of the old boxes, and it had ground to a dead halt, apparently failing to detect the CD on reboot. So that was out.

Into my hand slipped a CD marked "Xandros Linux Desktop 1.0". I inserted it, booted, and up came a nice install screen. Click, yes, agree, click, root password, yes take over entire disk, add one or two users, OK...

I installed the two old boxes and plugged them into the wall, and clicked Mozilla, fine. Clicked the File Manager and found myself in a fantastic file manager. "Not standard Linux, surely?" Clicked on the Xandros Networks icon and found something like Lindows' click'n'run, simple installation of packages. Underneath that, it's just apt-get, so when I opened a shell window and did "apt-get upgrade", 90 Mb of updates installed themselves onto the PC. Excellent!

The notebooks were a little more troublesome: the in-built network card just would not detect, some kind of IRQ conflict. I thought about installing a Windows just to determine the actual IRQ settings, but the idea just made me twitch like the mad inspector in the Pink Panther, and when I spotted a couple of unused network PCMCIA cards, I slapped them in and held my breath. Xandros detected them, reconfigured the network, and the notebooks were online.

What's great about Xandros?

1. It's Debian, which is the standard for excellent Linux distributions (so says my sysadmin, and he is never, ever wrong).

2. It has near flawless hardware detection, and does smart things with the information it gets. I did not have to change a single setting to get a fully working and almost optimally-tuned system. OK, I did go and change the color depth to 24bit instead of 16bit. That was _it_. I have never - ever - seen an OS that did such a good job, period.

3. Excellent file manager, just excellent.

4. Includes everything I wanted, and not a lot of extra junk.

5. Pretty, but not garish. Lindows, sorry, you should have chosen something else than XP as your role model. Xandros is classy and elegant, something you'd be proud to install for a client or friend.

6. Most administration and configuration works from the GUI. You can of course do the normal stuff from the command-line, but it seems kind of silly.

I did not try to Crossover office functions: apparently these let you install MS Office, use MSIE plugins in Mozilla, and let you install TrueType fonts. Honestly, we switched to OOo some time ago, and the fonts in Xandros look just fine.

There is one minor problem with Xandros, a problem shared with all Debian distributions: you have to be patient for new releases of software. For instance, I generally use OOo1.1, with its wonderfully useful "Export to PDF" feature. I could install this version on Xandros, but it would break the apt-get model. So I'm sitting with OOo1.0 and wondering whether I will crack and install OOo1.1 manually, or find some other way to produce PDFs on the one notebook that has to be able to do this.

Now the real problem with Xandros. When I installed it, I did not even realise it was commercial software. Sure, it has a license agreement, but does anyone actually read these? Certainly not on a Linux distro. Only when I went to the Internet to read about this obscure but interesting package did I realize that it was actually a $99 product, an inheritor of Corel Linux, which I'd installed in 1999 from a CD received from the Corel team themselves in the Bazaar expo in NYC.

In the meantime I've bought five copies of Xandros, which will arrive sometime in the next few days.

The problem: Xandros is a truly excellent product, and easily worth the $99 if you need MS Office support, or $40 if you can live without it. But it can only spread by word of mouth. And this means people installing it, trying it, and telling their friends about it. It was sheer luck that an (illegal) copy of this fell into my hands just when I needed it, but how can Xandros grow without some kind of instant gratification model?

Xandros should make the basic version (without the commercial software) available for free download. This is absolutely necessary, and it will not hurt their revenue one bit. Those of us who run OOo will admire Xandros and tell our friends and colleagues about it, and those who need MSOffice support will buy the commercial version. Why would you buy Crossover Office separately when it just means more installation work? No, get Xandros, it is all there for the same price and much fewer headaches.

Anyhow, Xandros took about one week to conquer our company: it provides what we've been wanting since we decided to migrate to Linux: a Debian based distro that is trivial to install and administrate, but is easy for naive users to work on. We can stop the business of boot-strapping Debian installs from diskette. We can stop worrying about what software people will install.

One of those old PCs is for my brother-in-law, who at 17, has only ever used a PC in cybershops, common in Africa and Europe. After a couple of days with Xandros, I asked him how it was going. He said, "nothing special, it works, sure".

That says it all.

This is a design proposal for a new kind of desktop metaphor.

The origin of this idea came from an article in the Economist that immediately struck a chord with me. Yes, yes, I thought as I read this. A tidy desk is indeed the sign of a twisted mind, if you're a creative person who works by distilling order from chaos.

Clutter is not bad, clutter is the raw material from which Good Things come. My filing system is my desk, all my active work spread out in concentric piles of interest. I can imagine my paleolithic stone-chipping ancestors spreading their tools and raw materials out on a ledge inside a dry warm cave...

If it works on my physical desk, why not on my screen? Why does every filing system available insist on classifying, organizing, and structuring my data? What if the very act of structuring my data was incompatible with the creative processes?

Why, after 20 years of using computers, do I keep all my really interesting and valuable stuff in a single folder called "temp"? Bizarre, I don't usually think of myself as disorganized or inefficient. Quite the contrary, you should see my kitchen: when people see it for the first time, they ask how long I was in the army for.

Why are UI designers making "smart menus" that hide unused options? Clearly I'm not the first person to think of a self-organizing UI. But while tweaking a broken model may be one way of improving things, I'm don't think it's useful.

No, the first rule of design is that if the user does not use a feature properly, the feature is badly designed. Ergo the conventional concept of a hierarchical filing system, which I don't use correctly unless I am forced to create structures for other people to use, is badly designed.

What I really want is a simple UI that works like my desk, or desks. I can throw stuff down, organize my ideas and projects into rough piles, and each new item causes the entire mess to reorganize so that the most useful, current, important things remain most obvious, and the least used, oldest, most ignored stuff gets relegated into an infinite background.

This journal entry will serve as prior art to protect this idea from some predatory corporation that thinks it can make a buck from it. Too late, buddies.

I've looked around, but no-one is working on anything like this AFAICS. So, I'm going to have to put my money where my mouth is, and implement this in software. When a prototype is ready, I'll announce it here.

So, a GSL example, hosted at iMatix.com, thanks guys.

The package is provided under the GPL for OSS developers and a commercial license for commercial developers (like RealiBase).

I've stuck in binaries for Win32 and for Linux, if the Linux binary don't work, you will want to pull down the RealiBase package from iMatix.com and rebuild that.

The definitions are in a bunch of DFL files, which stands for "database framework language", an XML language we use for defining relational databases. You'll see it lets you define tables, fields, indexes, etc. Fairly basic.

The main file is demo.dfl, which includes a bunch of other XML files (the "inclusion" is done by GSL at code generation time).

The main GSL script is dflmysql.gsl, and you invoke it like this:

  gsl -dfl:demo dflmysql

The result is a bunch of C and H files. Look at these, you'll see a series of functions and constants generated. OK, MySQL is trivial enough to access without a generated layer, but there are several advantages to doing it like this:

  1. You are guaranteed that accesses follow defined indexes.

  2. You always get data in the right type.

  3. No null columns in the database (you may like these, I don't).

  4. A full documentation of the database at all times (the DFL files).

  5. Generated SQL to create the tables.

The generated code is perfect: no bugs, full error checking, nothing left out. It's a neat example of why code generation is better than doing it by hand, in cases where you can.

Stick your questions here, I'll answer them all.