I don't know about European (DCF77) and US (WWV) transmitters, but the UK broadcast time (MSF) contains UT1-UTC corrections anyway. Not much equipment uses it, but it's there, so you could trivially build a clock that displays UT1. However, being based on astronomical observations, UT1 is inherently retrospective, which makes it hedged around with caveats --- there's UT1 in reality at the current second, which is unknowable, and UT1 at that second as seen from later, which is knowable.
Don't be silly. If you need monotonically increasing time, that's what TAI is: constant seconds, no leap seconds, ticked by atomic clocks.. If you need time that works for solar or celestial navigation because you want to sail boats using only a sextant, you use UT0 or UT1, so that the sun is in the right place relative to your watch, but you accept that seconds aren't constant: variations in the movement of the earth appear as variations in the length of seconds. UTC is a convenient compromise, with the constant seconds of TAI plus leap seconds to keep it within 0.9s of UT1. It's not good for long duration timing (leap seconds) and it's not good for accurate navigation (could be up to 0.9s), but it's the best compromise for civil time. The ITU-R are complaining that a timescale with properties X and Y doesn't have property Z, even though they could easily use timescale Z.
By the way, for extra fun, although all UK systems operate as if legal time is UTC, in fact it's GMT, which is either UT0 or UT1 depending on who you ask. There was legislation being worked on in 1997 to standardise on UTC, but it wasn't restarted after the change of government. So telecoms companies complaining that they don't like leap seconds but `have to' because of legal requirements are simply wrong: they should be ticking UT0 or UT1 for their billing systems, which don't have leap seconds anyway.
Commissioning a ring isn't hugely expensive: we bought two wedding rings and a an engagement ring, built to an agreed design as a set, for about five hundred quid. The stone in the engagement ring came from some family jewellery.
ian
Actually, LTSB verification involves being asked for (three, I think) letters from your password / passphrase. I believe that the operator has no access to the letters involved --- they are prompted to ask for three and eight, type them in, and now know what they are. If you don't know, they don't either: the letters aren't displayed to the operator. Online, you supply a username (which is related to you, not to your account) and password, and are then prompted for three characters from a passphrase as pull-down menu items (presumably to make key-loggers a little less useful). The telephone and online systems use different passphrases.
Now of course this isn't flawless: there are a lot of attacks one can envisage, mostly involving operators always asking for different letters --- ie if they already have three, five and eight, and are prompted to ask for three, five and nine, they ask for four, six and nine, supply three and five from their previous knowledge and now have six letters instead of the four they would otherwise have. By this technique they can get the password in n/3 attempts, less if (as is likely) you don't need all the letters to see what the whole word/phrase is. It's a thin attack given the chances of you arriving at the same operator, or the operator's confederate, that many times, but might be possible as a large conspiracy by a corrupt call centre (LTSB have in recent months re-on-shored all their call centres; make of that what you will). If you fail to authenticate, for whatever reason, you're asked for the same characters next time, so an attacker cannot make repeated attempts hoping to be asked for characters they already have if they don't get a favourable set the first time.
Some things about this story don't ring true, by the way. Firstly, LTSB have not, to my knowledge as a customer, had a limit on the length of pass phrases either for telephone banking or on-line banking as short as is claimed. The on-line `memorable information' (ie password) is six to fifteen characters, spaces not permitted, and I can't believe the voice system is different.
There are some things that could be improved. You can change the greeting between given name, given name plus surname and a few other options, but you can't have a custom greeting. That's a powerful phishing prevention mechanism: if I can customise my bank's website to greet me, after supplying my password but before supplying my selected characters from the passphrase, with a picture I supply (say) then that massively ups the problems a phisher faces. I have my passphrase as six random characters (ie knowledge of five doesn't provide the sixth) so that if I'm ever asked for character seven or greater I know something bad is happening, but it's not ideal. But the rest they do well: initial contact URL is https and won't work as http, ie http://online.lloydstsb.co.uk/ doesn't answer, so anyone bookmarking it will bookmark the https. Menus don't accept keyboard accelerators. More if I could think of it before my first coffee. I checked it through pretty thoroughly before signing the ts and cs, and I'm reasonably happy.
dtrace is great, but actually my experience as an administrator is that I use it less than I expect, because the kernel `just works'. I use it to attack badly behaved applications, but I've not used it for tuning anything like as much as I thought I would.
I run a mixed Linux and Solaris shop, but having replaced some Solaris boxes with Linux we're swinging massively back to Solaris 10.
ZFS, of course.
SMF. Being able to start services in a dependency tree is excellent if you have a multi-processor machine. And having services self-heal, including restarting any dependencies, is good for things like mail servers that use a lot of flakey milters.
FMA. Hardware self-healing (admittedly, this is essentially Sun hardware only, and in my experience better on Sparc than on AMD) is good.
Zones. Because sometimes full-blown virtualisation is too much like hard work.
Binary compatibility. I've got some SunOS 4.1.1 binaries from 1989, for which the source is long lost, running fine.
There's probably a Linux equivalent of rcapd, to limit the physical memory use of particular groups of processes, but I've never found one.
There's probably a Linux equivalent of processor sets, CPU shares and the Fair Shares Scheduler, but again I've never found one.
Horses for courses, but Solaris has much to offer even for shops that aren't traditionally tied to Sun. Hell, even my private ``1U box in someone else's datacentre'' server for my family is now a Solaris machine.
If your opponent is your email provider, someone who has a means to obtain your account password, or someone who has the means to obtain a warrant, SSL on the connection to your mailbox is worthless. In all of those scenarios, your opponent will obtain the data direct from your mailbox, no snooping of the connection required. Yes, it will protect against an active snooping attack, but those are less likely scenarios. In very, very general terms, you should ensure you have protected data at rest before you worry about data in motion.
And handily, if you PGP your mail (or S/MIME, if you must) that protects the data at rest _and_ the data in motion. The only benefit of using SSL then it to protect your password, and if you're not using either APOP or one of the MD5 IMAP authentication styles then indeed it's worth doing. But the PGP is numbers one, two and three on your list.
VMware Fusion mutters about supporting DirectX 9 on appropriate hardware, presumably for playing Windows games on Mac iron. It doesn't affect me: I use Fusion mostly to run Solaris on my Mac, plus a tiny Windows instance to support an X10 adapter and a Brother label printer, and I've not played a computer game since wasting a day on ADVENT in 1983. But I believe it's there for those that want it.
Williams Goldman's dictum on Hollywood's ability to spot both commercial and artistic winners: ``Nobody Knows Anything''.
Films don't by and large get their just deserts for decades. It's a popular game: take a great film. Look up the Oscars it got, or was nominated for (usually few). Look up the schlock that no-one remembers from that same year that got Oscars by the boatload. Try http://www.filmsite.org/noawards.html for the grim details.
And that doesn't even touch on "real" work, such as in-house app development.
In the businesses I'm familiar with, if all in-house applications development stopped tomorrow morning, the net effect would be massively positive. An IT function looking for work constantly changes reporting systems, so that no-one trusts the numbers (as they have no stable base) and maintains all their own reports (which are differently, but equally, wrong). There's no means to compare efficiency, quality, cost or anything over a period greater than six months, because the reports don't use consistent subsets of data which is changing its structure anyway. Reports written in year X can't be run against year X-1 or year X+1's data, because the underlying raw data isn't consistent either.
If all desktop and infrastructure development stopped stone dead the effects would be similar. People would, at last, be able to figure out how to do their jobs with the tools in front of them rather than either putting it off for the nirvana when the right tool turns up, or finding that their workflow is suddenly broken by the introduction of a new tool in place of one that worked. Most of the instability in networks comes from change, not maintenance, and most `improvements' aren't.
For most businesses, if you froze all the internal IT for twelve months or more and put all the resource into outward facing IT that actually, you know, impact customers (been there, done that, during a radical re-structuring) there would be only upside. No changes to reporting, HR systems, finance systems, payroll systems, ERP, mail servers, groupware...just focus on things that make money.
Parallel ports, PS/2 ports, and floppy disks were all declared "dead" a long time ago, but their corpses aren't being buried too quickly.
Floppies really are dead, though, in that not merely are no computers (for practical purposes) sold with them, the USB external ones are increasingly hard to come by, the media is disappearing from mainstream retail and although motherboards still have the connectors it's very rarely used. And that's because using floppies for long-term storage stopped about fifteen years ago, there hasn't been a capacity bump in twenty years and they are so small now that few datasets will now fit on one. And there's a perfect replacement: the USB stick.
But that's about the time scale: twenty years, and ten years from last practical use.
My theory is that it's an age thing. If you're forty or more, when you were at University you could trivially produce scenarios to cut code for where the problem was larger than the computing resources you had available. And for really quite small test cases, quick sorts really were faster than bubble sorts. And if you leaked memory, your shiny new code really did fall over. I've written code on systems with 128 bytes of RAM...
But if you're younger, the difference between O(n^2) and O(n log(n)) was just a curiosity. You've never written code on a system without not merely VM, but so much RAM that the VM aspect doesn't really matter. And you're used to edit-compile-go taking a few seconds, not a few days.
Which is great, and modern software environments are infinitely more productive than those of twenty five years ago. But when your website scales from a thousand records to a million records, or your code has to churn a thousand buffers per second for a year without being restarted so leaking one byte per operation just isn't acceptable, or it's running a 16-way machine flat-out and there's no end in sight, then you can't rely on hardware scale to bale you out.
And that's when having cut code on a pdp8 or an SC/MP or a Nascom comes into play...
The odds are greater that he is six degrees of separation from Saddam Hussein than any of the above.
Well. My father spent a few days in hospital after being hit by a bus (looking wrong way as he stepped out from another bus). As Head of IT I recently spent a few days in hospital after hitting a fence while cycling. I nearly walked under a car having had a few beers during a trip to install some servers in a country where they drive on the other side of the road.
And I'm trivially easily six degrees from Saddam Hussein, come to that. A friend's husband used to have regular meetings with Al Gore, so I reckon I can do Gore-UN_dudes-Hussein. My boss has had the occasional meeting with Blair, so that's Blair-UN_dudes-Hussein. Another friend sat opposite Clinton at a formal meal, so there's a route there. I had dinner a few years ago sat next to Scott McNealy, and I'd be surprised if I couldn't use him as the root of a variety of trees via California pols to US presidents to the UN to Hussein. RMS spent a week in my house in the eighties, and he's well enough travelled and connected that I bet I can do Stallman-some European politician-UN dudes-Hussein. Noted left-wing activist journalist Duncan Campbell (the `C' in the ABC trial) bought me a drink in a pub near Holborn tube station some years ago, and I bet I can get from him to arbitrary governments via at most one hop. A guy I was at school with has worked for the UN and is now a management consultant in the middle east with some time in Iraq under his belt: there must be a two-step link there. And so on, and so on.
Most of these links are through the US, and as I work in a provincial town in England I suspect that my six degrees are more restricted than a US local government employee.
Hand off DNS queries emerging from AD servers inside your firewall to caching-only servers in your DMZ. I have all my AD servers on RFC1918 IP numbers with no NAT, because they strike me as devices I'd prefer to keep as far away from the big bad Internet as possible.
That's true, but sometimes I think people on either side of the Atlantic fail to appreciate the radical differences in geography. For literally 99% of the UK population, they're no more than four or five miles from a telephone exchange (hence the 99%+ ADSL availability). Similarly mobile phones (hence coverage is for practical purposes continuous, excepting gaps a few hundred yards is size). That's because the UK is smaller than several US states, and the only areas which aren't densely populated are small patches in Wales and Northern England, plus the Scottish Highlands and Islands. About 40 million people live in an area less than two hundred miles square, although people say ``ah, Scotland's population density is much lower'' the central belt encompasses a similar proportion.
In mainland Europe the same's not true. France has a population less than the UK in twice the land mass, and there are substantial areas which are deeply rural. This is progressively more the case for Italy, Spain, etc, which are less urbanised. By the time you get to America, you have states like Oregon which are larger than the UK with FIVE PERCENT of the population.
ian
n France you get *at least* 50Mbps with fibre for about 30 per month, whis is about, what, $50, $55 ?
What, for people living up valleys in the Massif Central? People living in villages in the Dordogne? How's the fibre to the house at 50Mbps doing in Gavarnie?
Do you mean ``France'' or do you mean ``major metropolitan areas''?
A lot of people are reassured by a large range of products, even if they don't intend to buy them. Would you eat in a restaurant whose menu was only the things you wanted to eat that night? You might, but a lot of people like the reassurance that they could eat something else, if they wanted to, even if they don't. Similarly bookshops: those with extensive ranges make feel better about buying Harry Potter: they could buy Wittgenstein if they wanted, whereas at a shop that sells only Harry Potter they can't comfort themselves with that.
One of my guys went to a workshop last month run by VMware, looking at some of their new technologies. We (1000 desktops) were at the small end of the attendees: the rest of the people there were mostly from large corporates. The guy in charge wanted a quick straw poll on some issues, one of which was ``are you doing or planning to do Vista?''.
Seventy attendees. One hand.
A common reason for home upgrades is ``that's what I'm using at work / school / etc.''. As Vista has no traction in those markets, it's losing at home as well.
ian
Actually, the issue of Braun's involvement in building weapons is neither here nor there: if you regarded involvement in the Manhattan Project (which clearly killed more people than the V2) as a moral crime sufficient to render people unemployable, the US and the UK have no physics for about thirty years after the war.
But the use of slave labour in the Mittelwerk _is_ an issue, and something he should have been properly challenged over.
I did my O Levels at what was then a beacon comp, with an intake that many a grammar school would kill for. One (1) of my contemporaries got all As at O Level. Over in high-end selective schools, perhaps 5 or 10% of _their_ intake would manage the same feat. Today it's routine.
I did my A Levels at a technical college from which few conclusions can be drawn, but when I arrived at a redbrick University in 1983 I was dimly aware that people with three As, or maybe (whisper it) FOUR As existed over in the med school (I'm excluding General Studies, with good cause: my college didn't teach it, so I entered myself and got an A, so I presume it's worthless). No one on my Computer Science course had qualifications close to straight As, and this in an era where higher education uptake was about 10%.
Today, _everyone_ in a university like that, three or four times larger, will have qualifications `better' than mine. I don't know what sort of University A (computer science) B (physics) C (maths), or perhaps it's B (maths) and C (physics), I can't remember + A in general studies gets you into, but `not much' I think is the answer. With 40% take up as against 10%.
Looking at my kids' maths, at a heavily selective secondary school my daughter is several years behind where I was at the same age. At O level I did differential and integral calculus, with volumes of rotation a favourite sport, which is now optional at A Level.
Wars allow science to have the GDP of medium sized countries. It may not help the science, but it does help the engineering that makes it usable, and it's usable science that then drives the next generation.
Every morning on my way to work I pass where cavity magentrons were made into practical devices http://en.wikipedia.org/wiki/Cavity_magnetron and where the critical mass of Uranium was first deduced http://en.wikipedia.org/wiki/Frisch-Peierls_memorandum. The science didn't need huge budgets: the engineering that followed on from it did. A hour's drive takes me to Bletchley Park (http://en.wikipedia.org/wiki/Bletchley_park); again, the maths didn't need budgets, the engineering that followed did. Radar, atomic weapons and crypto: the spin offs drive a lot of the world today, but the raw science wouldn't have had as much influence without the money that science gives you.
If you want to duplicate their efforts with your own code that parallels community features, fine, burn yer money.
Part of the background to the GPL is the work RMS did re-implementing the features Symbolics put into the Lisp Machines so that LMI could release them. MIT had made available the sources and hardware designs for the Lisp Machines, and two companies --- Symbolics, who were well funded, and LMI, who were less so --- were producing commercial systems. The licenses said that changes made by the companies had to be given back to MIT, but that was not transitive: the changes didn't have to be made available to other licensees of the code. Symbolics refused to allow LMI to have the changes; that's commercially reasonable, ethically less so.
RMS decided to re-implement the Symbolics features and make the changes available to LMI. He had access to the source, but he just took the high-level specifications and interfaces and implemented against those. In 1978 or whenever, which was a kinder and gentler age, this was controversial enough. Today, if you took someone's copyright and GPL licensed material, read it, and then released a `re-implementation' against the same codebase, you'd need to have good lawyers.
But they can't incorporate the changes that are made by others into the non-GPL'd release, because those changes _are_ GPL'd.
You can dual-license your own code, of course, with one version released under the GPL and another version, derived from the same codebase, under a completely closed license. But the scenario in the grandparent was that a company could incorporate other users' changes back into the closed release. They can't: you can't dual-license, and then take changes made by other people in the GPL'd version and release them in the closed version.
Slashdot Mirror
ian
By the way, for extra fun, although all UK systems operate as if legal time is UTC, in fact it's GMT, which is either UT0 or UT1 depending on who you ask. There was legislation being worked on in 1997 to standardise on UTC, but it wasn't restarted after the change of government. So telecoms companies complaining that they don't like leap seconds but `have to' because of legal requirements are simply wrong: they should be ticking UT0 or UT1 for their billing systems, which don't have leap seconds anyway.
ian
Commissioning a ring isn't hugely expensive: we bought two wedding rings and a an engagement ring, built to an agreed design as a set, for about five hundred quid. The stone in the engagement ring came from some family jewellery. ian
Actually, LTSB verification involves being asked for (three, I think) letters from your password / passphrase. I believe that the operator has no access to the letters involved --- they are prompted to ask for three and eight, type them in, and now know what they are. If you don't know, they don't either: the letters aren't displayed to the operator. Online, you supply a username (which is related to you, not to your account) and password, and are then prompted for three characters from a passphrase as pull-down menu items (presumably to make key-loggers a little less useful). The telephone and online systems use different passphrases.
Now of course this isn't flawless: there are a lot of attacks one can envisage, mostly involving operators always asking for different letters --- ie if they already have three, five and eight, and are prompted to ask for three, five and nine, they ask for four, six and nine, supply three and five from their previous knowledge and now have six letters instead of the four they would otherwise have. By this technique they can get the password in n/3 attempts, less if (as is likely) you don't need all the letters to see what the whole word/phrase is. It's a thin attack given the chances of you arriving at the same operator, or the operator's confederate, that many times, but might be possible as a large conspiracy by a corrupt call centre (LTSB have in recent months re-on-shored all their call centres; make of that what you will). If you fail to authenticate, for whatever reason, you're asked for the same characters next time, so an attacker cannot make repeated attempts hoping to be asked for characters they already have if they don't get a favourable set the first time.
Some things about this story don't ring true, by the way. Firstly, LTSB have not, to my knowledge as a customer, had a limit on the length of pass phrases either for telephone banking or on-line banking as short as is claimed. The on-line `memorable information' (ie password) is six to fifteen characters, spaces not permitted, and I can't believe the voice system is different.
There are some things that could be improved. You can change the greeting between given name, given name plus surname and a few other options, but you can't have a custom greeting. That's a powerful phishing prevention mechanism: if I can customise my bank's website to greet me, after supplying my password but before supplying my selected characters from the passphrase, with a picture I supply (say) then that massively ups the problems a phisher faces. I have my passphrase as six random characters (ie knowledge of five doesn't provide the sixth) so that if I'm ever asked for character seven or greater I know something bad is happening, but it's not ideal. But the rest they do well: initial contact URL is https and won't work as http, ie http://online.lloydstsb.co.uk/ doesn't answer, so anyone bookmarking it will bookmark the https. Menus don't accept keyboard accelerators. More if I could think of it before my first coffee. I checked it through pretty thoroughly before signing the ts and cs, and I'm reasonably happy.
ian
On the ZFS point, just what filesystem do you think I put on my 20TB SAN?
dtrace is great, but actually my experience as an administrator is that I use it less than I expect, because the kernel `just works'. I use it to attack badly behaved applications, but I've not used it for tuning anything like as much as I thought I would.
Horses for courses, but Solaris has much to offer even for shops that aren't traditionally tied to Sun. Hell, even my private ``1U box in someone else's datacentre'' server for my family is now a Solaris machine.
ian
And handily, if you PGP your mail (or S/MIME, if you must) that protects the data at rest _and_ the data in motion. The only benefit of using SSL then it to protect your password, and if you're not using either APOP or one of the MD5 IMAP authentication styles then indeed it's worth doing. But the PGP is numbers one, two and three on your list.
ian
VMware Fusion mutters about supporting DirectX 9 on appropriate hardware, presumably for playing Windows games on Mac iron. It doesn't affect me: I use Fusion mostly to run Solaris on my Mac, plus a tiny Windows instance to support an X10 adapter and a Brother label printer, and I've not played a computer game since wasting a day on ADVENT in 1983. But I believe it's there for those that want it.
Williams Goldman's dictum on Hollywood's ability to spot both commercial and artistic winners: ``Nobody Knows Anything''. Films don't by and large get their just deserts for decades. It's a popular game: take a great film. Look up the Oscars it got, or was nominated for (usually few). Look up the schlock that no-one remembers from that same year that got Oscars by the boatload. Try http://www.filmsite.org/noawards.html for the grim details.
In the businesses I'm familiar with, if all in-house applications development stopped tomorrow morning, the net effect would be massively positive. An IT function looking for work constantly changes reporting systems, so that no-one trusts the numbers (as they have no stable base) and maintains all their own reports (which are differently, but equally, wrong). There's no means to compare efficiency, quality, cost or anything over a period greater than six months, because the reports don't use consistent subsets of data which is changing its structure anyway. Reports written in year X can't be run against year X-1 or year X+1's data, because the underlying raw data isn't consistent either.
If all desktop and infrastructure development stopped stone dead the effects would be similar. People would, at last, be able to figure out how to do their jobs with the tools in front of them rather than either putting it off for the nirvana when the right tool turns up, or finding that their workflow is suddenly broken by the introduction of a new tool in place of one that worked. Most of the instability in networks comes from change, not maintenance, and most `improvements' aren't.
For most businesses, if you froze all the internal IT for twelve months or more and put all the resource into outward facing IT that actually, you know, impact customers (been there, done that, during a radical re-structuring) there would be only upside. No changes to reporting, HR systems, finance systems, payroll systems, ERP, mail servers, groupware...just focus on things that make money.
ian
Floppies really are dead, though, in that not merely are no computers (for practical purposes) sold with them, the USB external ones are increasingly hard to come by, the media is disappearing from mainstream retail and although motherboards still have the connectors it's very rarely used. And that's because using floppies for long-term storage stopped about fifteen years ago, there hasn't been a capacity bump in twenty years and they are so small now that few datasets will now fit on one. And there's a perfect replacement: the USB stick.
But that's about the time scale: twenty years, and ten years from last practical use.
ian
But if you're younger, the difference between O(n^2) and O(n log(n)) was just a curiosity. You've never written code on a system without not merely VM, but so much RAM that the VM aspect doesn't really matter. And you're used to edit-compile-go taking a few seconds, not a few days.
Which is great, and modern software environments are infinitely more productive than those of twenty five years ago. But when your website scales from a thousand records to a million records, or your code has to churn a thousand buffers per second for a year without being restarted so leaking one byte per operation just isn't acceptable, or it's running a 16-way machine flat-out and there's no end in sight, then you can't rely on hardware scale to bale you out.
And that's when having cut code on a pdp8 or an SC/MP or a Nascom comes into play...
ian
Well. My father spent a few days in hospital after being hit by a bus (looking wrong way as he stepped out from another bus). As Head of IT I recently spent a few days in hospital after hitting a fence while cycling. I nearly walked under a car having had a few beers during a trip to install some servers in a country where they drive on the other side of the road.
And I'm trivially easily six degrees from Saddam Hussein, come to that. A friend's husband used to have regular meetings with Al Gore, so I reckon I can do Gore-UN_dudes-Hussein. My boss has had the occasional meeting with Blair, so that's Blair-UN_dudes-Hussein. Another friend sat opposite Clinton at a formal meal, so there's a route there. I had dinner a few years ago sat next to Scott McNealy, and I'd be surprised if I couldn't use him as the root of a variety of trees via California pols to US presidents to the UN to Hussein. RMS spent a week in my house in the eighties, and he's well enough travelled and connected that I bet I can do Stallman-some European politician-UN dudes-Hussein. Noted left-wing activist journalist Duncan Campbell (the `C' in the ABC trial) bought me a drink in a pub near Holborn tube station some years ago, and I bet I can get from him to arbitrary governments via at most one hop. A guy I was at school with has worked for the UN and is now a management consultant in the middle east with some time in Iraq under his belt: there must be a two-step link there. And so on, and so on.
Most of these links are through the US, and as I work in a provincial town in England I suspect that my six degrees are more restricted than a US local government employee.
ian
Hand off DNS queries emerging from AD servers inside your firewall to caching-only servers in your DMZ. I have all my AD servers on RFC1918 IP numbers with no NAT, because they strike me as devices I'd prefer to keep as far away from the big bad Internet as possible.
ian
That's true, but sometimes I think people on either side of the Atlantic fail to appreciate the radical differences in geography. For literally 99% of the UK population, they're no more than four or five miles from a telephone exchange (hence the 99%+ ADSL availability). Similarly mobile phones (hence coverage is for practical purposes continuous, excepting gaps a few hundred yards is size). That's because the UK is smaller than several US states, and the only areas which aren't densely populated are small patches in Wales and Northern England, plus the Scottish Highlands and Islands. About 40 million people live in an area less than two hundred miles square, although people say ``ah, Scotland's population density is much lower'' the central belt encompasses a similar proportion. In mainland Europe the same's not true. France has a population less than the UK in twice the land mass, and there are substantial areas which are deeply rural. This is progressively more the case for Italy, Spain, etc, which are less urbanised. By the time you get to America, you have states like Oregon which are larger than the UK with FIVE PERCENT of the population. ian
What, for people living up valleys in the Massif Central? People living in villages in the Dordogne? How's the fibre to the house at 50Mbps doing in Gavarnie?
Do you mean ``France'' or do you mean ``major metropolitan areas''?
ian
A lot of people are reassured by a large range of products, even if they don't intend to buy them. Would you eat in a restaurant whose menu was only the things you wanted to eat that night? You might, but a lot of people like the reassurance that they could eat something else, if they wanted to, even if they don't. Similarly bookshops: those with extensive ranges make feel better about buying Harry Potter: they could buy Wittgenstein if they wanted, whereas at a shop that sells only Harry Potter they can't comfort themselves with that.
One of my guys went to a workshop last month run by VMware, looking at some of their new technologies. We (1000 desktops) were at the small end of the attendees: the rest of the people there were mostly from large corporates. The guy in charge wanted a quick straw poll on some issues, one of which was ``are you doing or planning to do Vista?''. Seventy attendees. One hand. A common reason for home upgrades is ``that's what I'm using at work / school / etc.''. As Vista has no traction in those markets, it's losing at home as well. ian
Bruce gave a tribute to him from the stage at Cardiff last night. It's on the front page of his website http://www.brucespringsteen.net/news/index.html.
ian
Actually, the issue of Braun's involvement in building weapons is neither here nor there: if you regarded involvement in the Manhattan Project (which clearly killed more people than the V2) as a moral crime sufficient to render people unemployable, the US and the UK have no physics for about thirty years after the war. But the use of slave labour in the Mittelwerk _is_ an issue, and something he should have been properly challenged over.
I did my A Levels at a technical college from which few conclusions can be drawn, but when I arrived at a redbrick University in 1983 I was dimly aware that people with three As, or maybe (whisper it) FOUR As existed over in the med school (I'm excluding General Studies, with good cause: my college didn't teach it, so I entered myself and got an A, so I presume it's worthless). No one on my Computer Science course had qualifications close to straight As, and this in an era where higher education uptake was about 10%.
Today, _everyone_ in a university like that, three or four times larger, will have qualifications `better' than mine. I don't know what sort of University A (computer science) B (physics) C (maths), or perhaps it's B (maths) and C (physics), I can't remember + A in general studies gets you into, but `not much' I think is the answer. With 40% take up as against 10%.
Looking at my kids' maths, at a heavily selective secondary school my daughter is several years behind where I was at the same age. At O level I did differential and integral calculus, with volumes of rotation a favourite sport, which is now optional at A Level.
Every morning on my way to work I pass where cavity magentrons were made into practical devices http://en.wikipedia.org/wiki/Cavity_magnetron and where the critical mass of Uranium was first deduced http://en.wikipedia.org/wiki/Frisch-Peierls_memorandum. The science didn't need huge budgets: the engineering that followed on from it did. A hour's drive takes me to Bletchley Park (http://en.wikipedia.org/wiki/Bletchley_park); again, the maths didn't need budgets, the engineering that followed did. Radar, atomic weapons and crypto: the spin offs drive a lot of the world today, but the raw science wouldn't have had as much influence without the money that science gives you.
ian
RMS decided to re-implement the Symbolics features and make the changes available to LMI. He had access to the source, but he just took the high-level specifications and interfaces and implemented against those. In 1978 or whenever, which was a kinder and gentler age, this was controversial enough. Today, if you took someone's copyright and GPL licensed material, read it, and then released a `re-implementation' against the same codebase, you'd need to have good lawyers.
ian
You can dual-license your own code, of course, with one version released under the GPL and another version, derived from the same codebase, under a completely closed license. But the scenario in the grandparent was that a company could incorporate other users' changes back into the closed release. They can't: you can't dual-license, and then take changes made by other people in the GPL'd version and release them in the closed version.