No I don't, because there are _way_ too many confounding factors, and interdependent factors, for that kind of simplistic analysis. A _lot_ of things have changed in that time frame to affect crime trends. In the UK there are clear trends of some kinds of violence increasing and others decreasing. Determining the causal trends is very difficult, as is affecting them. Delays of up to 20-40 years between peaks in causes and effects is plausible, making it even harder.
And Doom isn't really violent by modern standards, is it?:-)
This kind of stuff can be studied and analysed, but it's difficult and requires a lot of care. Sweeping moral generalisations (either way) are nothing more than wishful thinking. Unfortunately they are "obvious" to many people despite being logically flawed arguments, which is why so many people believe in one position or the other (the one they personally like of course). Not that anyone is to be blamed; social perspective and policy are often a battle of ideas, rather than deep analysis. And that's probably for the best.
The important question, in that case, is whether using a pedophilia simulator make for more attacks in the real world, or less. I don't think the answer is known at this point... Some people say (as with violence in general) that the urge is dissipated by games, or even passive media like TV. Some people say the opposite. There are strong opinions either way, but I have yet to read of persuasive studies with conclusive results. Not even the conclusion "has no effect".
Let's say your file says the equivalent of "You may use this file under the terms of the GPL (v2 or later), LGPL or MPL, at your option". (Mozilla source is triple-licensed like this, I believe).
Then you want to add code from somebody else's project to your file. Their file says "You may use this file under the terms of the GPL v2".
When you add their code to your file, if you don't change the licensing at the start of your file, and you publish the new version, you are telling everyone that they are allowed to use the combination under any of the licenses which you listed.
But you don't have permission to do that.
The code you used from someone else may only be used under the terms of GPL v2 (in this example). And the GPL v2 does not allow the code to be re-licensed under any of those other license. (If it was BSD instead, you could do that).
Therefore, you must change the license of your file to say GPL v2 only, or not incorporate their code. (Or you can just not publish the combination - you are free to combine them and use the combination without distributing the result to anyone.)
I think you're mistaken about Linus' model. Some patches, he reviews himself, but many others he doesn't review: they are simply pulled from other people he trusts, as a time to suit him. This is what 'lieutenants' are for.
If you're thinking of a smaller project where _everyone_ is trusted, then it's quite similar to the centralised model anyway: patches will be pulled between developers without review, because they trust each other.
The one major difference is that individual published trees will have been checked by the person publishing them to whatever level of quality they care about. In a small team, it might be "the project compiles". (This avoids the problem which occurs with centralised scms and large numbers of developers: if everyone only rarely checks in a patch which breaks the build, the build will be broken a lot of the time because of the large number of developers.)
That is again very similar to the centralised process with release management: committing something to a published tree is analagous to a minor release.
When you want to get the most up to date code, including code that's not gone through the basic checks of one particular developer (or testing script), you can: you simply pull directly from all the people who might have some changes of interest. In the case of a small team where everyone is trusted, that means you pull from everyone.
If that's not convenient enough, you can set up a central place which constantly pulls or is pushed to by team members. Then it's quite like centralised development - except you can set up multiple such places with the work of different overlapping teams. That's quite useful when you have to work with third parties where you might trust them for some things but need to review their work for others in the same project.
If you look at OSS projects which use centralised SCM tools, you'll often see that there are multiple repositories in effect, even though they're not called that. Individual contributors have their own copy of code, make local modifications, and _eventually_ they are applied centrally, after some kind of review. That is, only a few trusted develops have real check-in privileges, so everyone else (i.e. casual developers) makes patches and posts them to mailing lists, and re-post after each merge or minor change until acceptance. For contributors who maintain multiple patches for a while, until they are accepted, the contributors are doing distributed SCM "by hand". Sometimes that is quite hard work. Distributed tools help.
(In my experience, commercial embedded systems projects are exactly the same except the communications are much less efficient: often, downstream customers patch their code locally, and have to maintain their local patches with changing upstream versions, but they have no good way to send the patches upstream efficiently due to bureaucracy, and new upstream versions tend to come as a gigantic patch which is harder to merge with local changes than a series of smaller patches.)
For sure. I'm male, and I have enjoyed many sexual experiences where I didn't orgasm and my partner (male or female) did, and I'm quite happy with that. Also there are 'energy orgasms' which are much more of a mental experience, and to be recommended.
If tuned low frequency fields affect cell division and geometry - I find myself wondering if the body _uses_ this effect to coordinate the growth of structures...
Harmonics are a multiple of the original frequency, i.e. higher. So cellphone harmonics would be higher than a few Ghz. However, radios are tuned to minimise emission of harmonics anyway.
But on a related note (get it;-), cellphones emit a bunch of lower frequencies as well, but not intentionally part of the radio signal. E.g. just listen to the way you can hear the cellphone transmission pattern when it's near speakers.
There's also non-linear effects, where the modulated GHz field interacts non-linearly with some material (the phone, your body, etc.) to produce lower frequencies at much lower intensity than the original.
Sweep your eye across the screen at a speed of one pixel per refresh, looking at a region of uniform colour such as a window's title bar. When you get the speed just right, on a colour which is not one of the 64-bit exact values, the dithering pattern becomes apparent, and seems to move across the screen at that speed, superimposed over the main image.
At least, it does for me. (Or did. I haven't looked at low quality panels in a while).
This is because the time part of the temporal-spatial dither pattern is effectively cancelled by the eye motion, and your eyes (if they are sensitive to it) then respond the spatial dither pattern.
Even from a distance, although the average colour is the same, the spatial dither pattern adds a high frequency component to the image which is normally obscured by the temporal dithering. If you are close enough, a chess-board pattern may be visible - seeming to move across the screen. But even from a distance (or not as focused), the quality of the colour changes somehow.
I found it quite hard to avoid, and quite disconcerting, in a shop selling cheap LCD monitors. One minute I was looking at a pretty picture on the screen. The next, I moved my eyes to look at something else, and from time to time the colours on the screen took on the quality associated with the dither pattern - only when sweeping my eyes at just the right speed. My old laptop did the same. My current laptop doesn't - if it's dithering, it's much subtler than the old one.
I did that too. As someone else said, I seem to be an 'auditory thinker' - at least when reading. (I'm visual-spatial when doing abstract reasoning:-)
But I gave it a chance, and noticed that it's because I'm used to those kind of line breaks meaning 'pause' or 'breath' in common formatting. Including poetry, and block text where line breaks are similar to paragraphs.
When I scanned it again a few times, I found it gets easier (for me) to scan it smoothly.
Then I found it was tiring, because my eye has to sweep vertically faster than usual. About the same speed as reading vertical text. (Curiously, the vertical sweep felt it quite different from scanning text quickly - which I do a lot as a programmer).
But that might be something the eye muscles can adapt to.
All that said, I've always been good at adapting quickly to unusual writing forms, like reading upside-down or rotated. So maybe I'm not representative of other people who found the formatted version more difficult.
Similar here. 1GB RAM. 2GHz per core according to/proc/cpuinfo when busy. (Yet strangely Gnome CPU Frequency Scaling Monitor shows 2GHz (100%) and 1GHz (50%), at the same time as cpuinfo shows 2GHz for both).
Shutdown time may well be comparable with 20 seconds (though it feels longer, I haven't measured).
I boot with text, because I have encrypted/home and text is needed to enter the passphrase. Each little bit of the boot process is quite fast, except for the long delays at things I mentioned and a few others. The slow bits add up, times clearly shown in/var/log/boot, which is where I get 75 seconds.
Ubuntu Feisty boot time to usable desktop: 95 seconds.
Well I'm using Ubuntu Feisty, presumably with Upstart, and it takes about 75 seconds to boot to the Gnome login screen, on my Core Duo laptop.
The kernel is only a tiny fraction of that; the kernel starts very quickly.
Most of it is surprising delays in some of the boot scripts. For example "Starting system message bus dbus" is followed by a 15 second delay before the next message, "Starting Hardware abstraction layer hald". "Starting kernel event manager" is followed by a 5 second delay before the next message, "Loading hardware drivers". "Assembling MD array mdarrays" takes 10 seconds before the next thing, "Setting up LVM Volume Groups". (Surely MD setup should be near-instant, especially as there are no MD arrays? And I shouldn't have to tweak it manually to get rid of that step.) Most things take about a second. They add up to about 75 seconds.
Then, after logging in at the Gnome login screen, if the cache is cold following a boot, it takes about 20 seconds to start all the applets, before it's possible to use the menu or icons. The panel itself appears quite quickly, but it's not usable until everything has appeared on it.
Shutting down is pretty slow too. I usually do Control-Alt-SysRq-S-U-B/H if I want to power off completely or reboot, to avoid the shutting down time.
I'd like to refute your argument that a "God who sits within the universe" is constrained to be just one more observer in the relativistic framework.
Even a single human being does not exist in a single inertial reference frame.
Our bodies are composed of many parts; our brains are. The parts are separated over a region of space. They have different motions, and thus each have their own reference frame. There is something common, coherent about the parts together of course. We can reasonably talk about the reference frame of a person, and accurately due to the numerics involved, but that's not an exact description.
We can imagine an intelligent entity (person, machine, whatever) spanning a larger region of space. Perhaps even a very large region, like a huge brain, or a cluster of them communicating over a network. An interplanetary artifical intelligence? It's not impossible. Arguably, we humans are just clusters of communicating cells; why not something like that on a large scale? What's the inertial reference frame of an intelligent entity like that? Again, we can approximate, using the centre of mass perhaps.
What about an artificial intelligence consisting of communicating nodes orbiting, some distance from each other, around a black hole? It gets tough to reason even approximately about "the" inertial reference frame of the intelligent entity then.
What about something like that _really_ large?
Consider, if you will, an intelligent being whose physical basis spans the entire physical universe. Possibly the parts are smaller sub-intelligences (all the way down to us?) but that's not necessary; it could be alongside us. Basically a universe-sized brain/body. That would "sit within the universe", and be constrained by relativity, limiting both what it can think (at that scale, relativity _is_ an important constraint computationally), and what it can observe or affect physically. Yet it would be different than "just" another observer like you and I.
Yes it is. It's also perfectly normal to be astonished at the idea of having a girlfriend at age 11. (It's equally normal to be unsurprised - there's plenty of variation.) When I was at school, pretty much nobody had a girlfriend at that age.
No: the contention ratios (50:1 and 20:1 typically) advertised in the UK weren't for the internet, they were for BT Wholesale's ATM network from your ADSL modem at the telephone exchange, to the ISP's relatively few site(s) where they connect to the internet backbone (and keep the web caches, mail servers etc.). Nearly all ISPs had to use BT Wholesale, because only BT could fit the necessary equipment at the telephone exchange.
The ISP's capacity to the internet was almost never advertised, and the only way to get an idea of the capacity offered was reviews and ratings by others customers.
It's worth bearing in mind that there is no magic 'capacity to the internet' figure anyway. The internet is a fabric. You might have a 100Gbit/s pipe to a tier 1 provider, but that doesn't guarantee you 100Gbit/s to every other node in the world with a similar pipe - it will vary to different places in the world. So what's the real capacity? All you can do is measure and predict, and maintain statistical guarantees; same with latency, and other figures.
I use Ubuntu now (and used to use Red Hat). I'm quite happy with it. But to be fair to the GP, when I installed Ubuntu Dapper from a CD (back when Dapper was the current stable thing to use), the install process turned into a black screen somewhere before the end, when it was trying to configure X. And this is on a laptop with Intel graphics chipset, one of the best supported. So I have to agree that Ubuntu does have hardware problems from time to time, just like all the other distros.
Since getting past that, though, Ubuntu has been pretty good. Oh, the wireless has locked up the whole system from time to time, boot has locked up with ACPI errors, sound has stopped working after kernel upgrades, kernel upgrades have sometimes failed to update the Grub boot menu, and it's impossible to switch X from the laptop screen to an external screen without restarting X. But apart from those niggles, it's been pretty good:-)
Ah, Color Kinetics, the folks who patented pulse-width-modulation for varying LED colours so other LED manufacturers can't do it or have to pay through the nose. I wouldn't mind, but it's such an obvious technique.
I've been tossing around an idea of an anti-captcha, though. Throw in a captcha, and right below it, have a note that says "now, disregard the above captcha and type 'notabot' in the box". I'll probably implement it to see what happens.
Obviously that will only work until the spammers add a rule to check for what you're doing. If your method remains a minority, they probably won't bother.
Given that, why bother asking the user to type anything or show them the graphic? Just use Javascript to enter the text into a hidden field and hide the image ("visibility: none") in a way that looks to a bot like it's shown. Thus, the users aren't bothered but you'll hopefully catch the (current) bots.
Once you insert an ORDER BY clause into your query, the DB automatically has to process all the data before it can send you any results.
That's not true, if the ORDER BY column(s) are ordered-indexed, or if it/they are the table's primary storage ordering. Which one would hope applies in a query like "select most recent 5 articles" - if they are stored by date, and any WHERE clause is not too sparse a filter.
Think about traversing a B-tree, or any kind of tree. You don't have to read the whole tree to begin streaming out the smallest elements.
What's even worse then that are the admins without a clue that silently drop e-mail.
That's a problem. But what of the alternatives? Notifying the recipient is pointless; they'll receive 6000 spam notifications a day instead of spams. Notifying the sender is done by some mail systems, and that's as much of a problem as spam: I have received 5000 "backscatter" messages a day for several months, because of spams which use one of my domains as the sender address. I can (and do) filter those, but then I wouldn't be notified if a mail that I really did send was dropped.
The only technical solution I see is a combination:
strong authentication of sender addresses;
returning dropped messages including the sender authentication so unwanted backscatter can be dropped but wanted is read by the legitimate sender;
whitelisting at the recipient (if I've accepted mail from you before, I probably accept future ones too), with good user interface (i.e. one click or keypress) to white/blacklist senders, and expiring entries if nothing is exchanged for a long time
ubiquitous and socially accepted challenge-response for senders not yet whitelisted, again requiring a good user interface, not obscure and tedious and easily lost recipient-specific instructions each time (this goes well with point 2)
It's a major upheaval, difficult because it would require a lot of cooperation, and consensus which we don't have, and I'm still not sure it would work. But something along those lines might work about as well as web sites that require registration to avoid comment spam - the requirement for "real" user registration blocks comment spam moderately well.
540kHz is the bottom end of medium wave (MW). Other AM radio sets, especially older or more "international" sets include long wave (LW), which goes down to about 150kHz.
Also, as your assumptions are incorrect, microwave radiation's focus is primarily in a line-of-sight fashion although the omni antennas diffuse this outward to maximize reception in handsets.
All EM radiation is line of sight for a sufficiently wide beamwidth and sufficiently wide obstacles.:)
We characterise 30cm microwave radiation as "line-of-sight" for practical transmission because 30cm is small compared with buildings, trees, walls, windows, doors, and other typical obstacles in the environment.
Compare with, say, AM radio which has wavelengths up to 2km at the extreme. Such long waves are barely affected by individual buildings and trees (unless you're inside a conductive building) because of the scale difference - effectively, those waves travel around building-sized obstacles, but are blocked by large structures such as mountains. So we don't describe AM radio as line-of-sight for that reason. (But if you have a space probe and the moon gets in the way, those transmissions are line-of-sight at that scale. It's all relative).
But it is not accurate to think of 30cm microwaves as "line-of-sight" at the scale of a few centimetres - the distance from the cellphone antenna to the LEDs.
A cellphone antenna is omnidirectional largely because you cannot emit a narrow, highly directional beam of 30cm waves from an antenna that fits inside a cellphone.
With current signal processing abilities, if it were possible to emit a highly directional beam (or the equivalent when you take into account multipaths from reflections etc.; that doesn't affect this argument, it just generalises the maths) that would be a huge improvement in spectrum usage, because many phones could use the same frequencies concurrently in the same cell. Just imagine a pencil beam from the base station to each handset - they'd each have a private channel and there would be no interference. But it's not possible because of the physics and the size of handsets relative to the wavelength. It is possible at the base station only, and that's the reason base stations have many antennae pointing in different directions.
It is possible with larger transcievers - and it is done now by MIMO transcievers appearing in current wireless access points - but you'll notice they have at least two antennae, spaced apart (which is essential), and also the wavelength is smaller (about 10cm or 5cm).
(It is actually possible to produce a highly directional beam when several small transcievers cooperate. Through cooperation it is also possible to get a similar increase in data rate for each transciever, as if each one had a directional connection to the base station - a technique I call "distributed lensing". But currently available kit doesn't do that, as far as I know. I'm sure it will, it's the obvious next step after MIMO.)
Back to the point...
On the question of amount of power absorbed: I think we can safely say the power absorbed by the LEDs is, well, the amount of power needed to light a modern blinking LED. Which, as you say, is a few mW at most.:)
On the analogy with a boat: you need large waves (which you don't get from a throwable rock) from an emitter smaller than the waves (say, a vertical pipe with outlet just below the water surface, pumping water in and out), and a tiny boat that _doesn't_ bob freely; it must be anchored with a tuned, damped spring to make a good energy absorber.
(And if you really meant 359.9 degrees not blocked by the boat, the boat is really far away from the emitter.)
But you make a good point anyway;) Only a fraction of the energy of the circular wave will be absorbed even if the boat is a perfect absorber, because as you note most of the wave energy goes elsewhere.
In the case of large waves from a small transmitter and a good but small absorber close to the transmitter, that is still true: there's a maximum amount which can be absorbed by a perfect absorber, and yes i
Slashdot Mirror
No I don't, because there are _way_ too many confounding factors, and interdependent factors, for that kind of simplistic analysis. A _lot_ of things have changed in that time frame to affect crime trends. In the UK there are clear trends of some kinds of violence increasing and others decreasing. Determining the causal trends is very difficult, as is affecting them. Delays of up to 20-40 years between peaks in causes and effects is plausible, making it even harder.
:-)
And Doom isn't really violent by modern standards, is it?
This kind of stuff can be studied and analysed, but it's difficult and requires a lot of care. Sweeping moral generalisations (either way) are nothing more than wishful thinking. Unfortunately they are "obvious" to many people despite being logically flawed arguments, which is why so many people believe in one position or the other (the one they personally like of course). Not that anyone is to be blamed; social perspective and policy are often a battle of ideas, rather than deep analysis. And that's probably for the best.
The important question, in that case, is whether using a pedophilia simulator make for more attacks in the real world, or less. I don't think the answer is known at this point... Some people say (as with violence in general) that the urge is dissipated by games, or even passive media like TV. Some people say the opposite. There are strong opinions either way, but I have yet to read of persuasive studies with conclusive results. Not even the conclusion "has no effect".
I'll explain with an example.
Let's say your file says the equivalent of "You may use this file under the terms of the GPL (v2 or later), LGPL or MPL, at your option". (Mozilla source is triple-licensed like this, I believe).
Then you want to add code from somebody else's project to your file. Their file says "You may use this file under the terms of the GPL v2".
When you add their code to your file, if you don't change the licensing at the start of your file, and you publish the new version, you are telling everyone that they are allowed to use the combination under any of the licenses which you listed.
But you don't have permission to do that.
The code you used from someone else may only be used under the terms of GPL v2 (in this example). And the GPL v2 does not allow the code to be re-licensed under any of those other license. (If it was BSD instead, you could do that).
Therefore, you must change the license of your file to say GPL v2 only, or not incorporate their code. (Or you can just not publish the combination - you are free to combine them and use the combination without distributing the result to anyone.)
Some people are lucky. On bad days, I get a spam every 15 seconds on average. On good days, it's every 80 seconds.
I think you're mistaken about Linus' model. Some patches, he reviews himself, but many others he doesn't review: they are simply pulled from other people he trusts, as a time to suit him. This is what 'lieutenants' are for. If you're thinking of a smaller project where _everyone_ is trusted, then it's quite similar to the centralised model anyway: patches will be pulled between developers without review, because they trust each other. The one major difference is that individual published trees will have been checked by the person publishing them to whatever level of quality they care about. In a small team, it might be "the project compiles". (This avoids the problem which occurs with centralised scms and large numbers of developers: if everyone only rarely checks in a patch which breaks the build, the build will be broken a lot of the time because of the large number of developers.) That is again very similar to the centralised process with release management: committing something to a published tree is analagous to a minor release. When you want to get the most up to date code, including code that's not gone through the basic checks of one particular developer (or testing script), you can: you simply pull directly from all the people who might have some changes of interest. In the case of a small team where everyone is trusted, that means you pull from everyone. If that's not convenient enough, you can set up a central place which constantly pulls or is pushed to by team members. Then it's quite like centralised development - except you can set up multiple such places with the work of different overlapping teams. That's quite useful when you have to work with third parties where you might trust them for some things but need to review their work for others in the same project. If you look at OSS projects which use centralised SCM tools, you'll often see that there are multiple repositories in effect, even though they're not called that. Individual contributors have their own copy of code, make local modifications, and _eventually_ they are applied centrally, after some kind of review. That is, only a few trusted develops have real check-in privileges, so everyone else (i.e. casual developers) makes patches and posts them to mailing lists, and re-post after each merge or minor change until acceptance. For contributors who maintain multiple patches for a while, until they are accepted, the contributors are doing distributed SCM "by hand". Sometimes that is quite hard work. Distributed tools help. (In my experience, commercial embedded systems projects are exactly the same except the communications are much less efficient: often, downstream customers patch their code locally, and have to maintain their local patches with changing upstream versions, but they have no good way to send the patches upstream efficiently due to bureaucracy, and new upstream versions tend to come as a gigantic patch which is harder to merge with local changes than a series of smaller patches.)
For sure. I'm male, and I have enjoyed many sexual experiences where I didn't orgasm and my partner (male or female) did, and I'm quite happy with that. Also there are 'energy orgasms' which are much more of a mental experience, and to be recommended.
If tuned low frequency fields affect cell division and geometry - I find myself wondering if the body _uses_ this effect to coordinate the growth of structures...
Harmonics are a multiple of the original frequency, i.e. higher. So cellphone harmonics would be higher than a few Ghz. However, radios are tuned to minimise emission of harmonics anyway. But on a related note (get it ;-), cellphones emit a bunch of lower frequencies as well, but not intentionally part of the radio signal. E.g. just listen to the way you can hear the cellphone transmission pattern when it's near speakers.
There's also non-linear effects, where the modulated GHz field interacts non-linearly with some material (the phone, your body, etc.) to produce lower frequencies at much lower intensity than the original.
I haven't tried it, but KLone looks interesting.
The number of comparisons is different from, and larger than, the number of page accesses.
Sweep your eye across the screen at a speed of one pixel per refresh, looking at a region of uniform colour such as a window's title bar. When you get the speed just right, on a colour which is not one of the 64-bit exact values, the dithering pattern becomes apparent, and seems to move across the screen at that speed, superimposed over the main image. At least, it does for me. (Or did. I haven't looked at low quality panels in a while). This is because the time part of the temporal-spatial dither pattern is effectively cancelled by the eye motion, and your eyes (if they are sensitive to it) then respond the spatial dither pattern. Even from a distance, although the average colour is the same, the spatial dither pattern adds a high frequency component to the image which is normally obscured by the temporal dithering. If you are close enough, a chess-board pattern may be visible - seeming to move across the screen. But even from a distance (or not as focused), the quality of the colour changes somehow. I found it quite hard to avoid, and quite disconcerting, in a shop selling cheap LCD monitors. One minute I was looking at a pretty picture on the screen. The next, I moved my eyes to look at something else, and from time to time the colours on the screen took on the quality associated with the dither pattern - only when sweeping my eyes at just the right speed. My old laptop did the same. My current laptop doesn't - if it's dithering, it's much subtler than the old one.
I did that too. As someone else said, I seem to be an 'auditory thinker' - at least when reading. (I'm visual-spatial when doing abstract reasoning :-)
But I gave it a chance, and noticed that it's because I'm used to those kind of line breaks meaning 'pause' or 'breath' in common formatting. Including poetry, and block text where line breaks are similar to paragraphs.
When I scanned it again a few times, I found it gets easier (for me) to scan it smoothly.
Then I found it was tiring, because my eye has to sweep vertically faster than usual. About the same speed as reading vertical text. (Curiously, the vertical sweep felt it quite different from scanning text quickly - which I do a lot as a programmer).
But that might be something the eye muscles can adapt to.
All that said, I've always been good at adapting quickly to unusual writing forms, like reading upside-down or rotated. So maybe I'm not representative of other people who found the formatted version more difficult.
Oops, mis-clicked a wrongful moderation there; replying just to undo the mod. Ignore me.
Similar here. 1GB RAM. 2GHz per core according to /proc/cpuinfo when busy. (Yet strangely Gnome CPU Frequency Scaling Monitor shows 2GHz (100%) and 1GHz (50%), at the same time as cpuinfo shows 2GHz for both).
Shutdown time may well be comparable with 20 seconds (though it feels longer, I haven't measured).
I boot with text, because I have encrypted /home and text is needed to enter the passphrase. Each little bit of the boot process is quite fast, except for the long delays at things I mentioned and a few others. The slow bits add up, times clearly shown in /var/log/boot, which is where I get 75 seconds.
Ubuntu Feisty boot time to usable desktop: 95 seconds.
Well I'm using Ubuntu Feisty, presumably with Upstart, and it takes about 75 seconds to boot to the Gnome login screen, on my Core Duo laptop.
The kernel is only a tiny fraction of that; the kernel starts very quickly.
Most of it is surprising delays in some of the boot scripts. For example "Starting system message bus dbus" is followed by a 15 second delay before the next message, "Starting Hardware abstraction layer hald". "Starting kernel event manager" is followed by a 5 second delay before the next message, "Loading hardware drivers". "Assembling MD array mdarrays" takes 10 seconds before the next thing, "Setting up LVM Volume Groups". (Surely MD setup should be near-instant, especially as there are no MD arrays? And I shouldn't have to tweak it manually to get rid of that step.) Most things take about a second. They add up to about 75 seconds.
Then, after logging in at the Gnome login screen, if the cache is cold following a boot, it takes about 20 seconds to start all the applets, before it's possible to use the menu or icons. The panel itself appears quite quickly, but it's not usable until everything has appeared on it.
Shutting down is pretty slow too. I usually do Control-Alt-SysRq-S-U-B/H if I want to power off completely or reboot, to avoid the shutting down time.
-- Jamie
I'd like to refute your argument that a "God who sits within the universe" is constrained to be just one more observer in the relativistic framework.
Even a single human being does not exist in a single inertial reference frame.
Our bodies are composed of many parts; our brains are. The parts are separated over a region of space. They have different motions, and thus each have their own reference frame. There is something common, coherent about the parts together of course. We can reasonably talk about the reference frame of a person, and accurately due to the numerics involved, but that's not an exact description.
We can imagine an intelligent entity (person, machine, whatever) spanning a larger region of space. Perhaps even a very large region, like a huge brain, or a cluster of them communicating over a network. An interplanetary artifical intelligence? It's not impossible. Arguably, we humans are just clusters of communicating cells; why not something like that on a large scale? What's the inertial reference frame of an intelligent entity like that? Again, we can approximate, using the centre of mass perhaps.
What about an artificial intelligence consisting of communicating nodes orbiting, some distance from each other, around a black hole? It gets tough to reason even approximately about "the" inertial reference frame of the intelligent entity then.
What about something like that _really_ large?
Consider, if you will, an intelligent being whose physical basis spans the entire physical universe. Possibly the parts are smaller sub-intelligences (all the way down to us?) but that's not necessary; it could be alongside us. Basically a universe-sized brain/body. That would "sit within the universe", and be constrained by relativity, limiting both what it can think (at that scale, relativity _is_ an important constraint computationally), and what it can observe or affect physically. Yet it would be different than "just" another observer like you and I.
Just a little thought.
-- Jamie
Yes it is. It's also perfectly normal to be astonished at the idea of having a girlfriend at age 11. (It's equally normal to be unsurprised - there's plenty of variation.) When I was at school, pretty much nobody had a girlfriend at that age.
No: the contention ratios (50:1 and 20:1 typically) advertised in the UK weren't for the internet, they were for BT Wholesale's ATM network from your ADSL modem at the telephone exchange, to the ISP's relatively few site(s) where they connect to the internet backbone (and keep the web caches, mail servers etc.). Nearly all ISPs had to use BT Wholesale, because only BT could fit the necessary equipment at the telephone exchange.
The ISP's capacity to the internet was almost never advertised, and the only way to get an idea of the capacity offered was reviews and ratings by others customers.
It's worth bearing in mind that there is no magic 'capacity to the internet' figure anyway. The internet is a fabric. You might have a 100Gbit/s pipe to a tier 1 provider, but that doesn't guarantee you 100Gbit/s to every other node in the world with a similar pipe - it will vary to different places in the world. So what's the real capacity? All you can do is measure and predict, and maintain statistical guarantees; same with latency, and other figures.
I use Ubuntu now (and used to use Red Hat). I'm quite happy with it. But to be fair to the GP, when I installed Ubuntu Dapper from a CD (back when Dapper was the current stable thing to use), the install process turned into a black screen somewhere before the end, when it was trying to configure X. And this is on a laptop with Intel graphics chipset, one of the best supported. So I have to agree that Ubuntu does have hardware problems from time to time, just like all the other distros.
:-)
Since getting past that, though, Ubuntu has been pretty good. Oh, the wireless has locked up the whole system from time to time, boot has locked up with ACPI errors, sound has stopped working after kernel upgrades, kernel upgrades have sometimes failed to update the Grub boot menu, and it's impossible to switch X from the laptop screen to an external screen without restarting X. But apart from those niggles, it's been pretty good
Ah, Color Kinetics, the folks who patented pulse-width-modulation for varying LED colours so other LED manufacturers can't do it or have to pay through the nose. I wouldn't mind, but it's such an obvious technique.
Obviously that will only work until the spammers add a rule to check for what you're doing. If your method remains a minority, they probably won't bother.
Given that, why bother asking the user to type anything or show them the graphic? Just use Javascript to enter the text into a hidden field and hide the image ("visibility: none") in a way that looks to a bot like it's shown. Thus, the users aren't bothered but you'll hopefully catch the (current) bots.
-- Jamie
Once you insert an ORDER BY clause into your query, the DB automatically has to process all the data before it can send you any results.
That's not true, if the ORDER BY column(s) are ordered-indexed, or if it/they are the table's primary storage ordering. Which one would hope applies in a query like "select most recent 5 articles" - if they are stored by date, and any WHERE clause is not too sparse a filter.
Think about traversing a B-tree, or any kind of tree. You don't have to read the whole tree to begin streaming out the smallest elements.
That's a problem. But what of the alternatives? Notifying the recipient is pointless; they'll receive 6000 spam notifications a day instead of spams. Notifying the sender is done by some mail systems, and that's as much of a problem as spam: I have received 5000 "backscatter" messages a day for several months, because of spams which use one of my domains as the sender address. I can (and do) filter those, but then I wouldn't be notified if a mail that I really did send was dropped.
The only technical solution I see is a combination:
It's a major upheaval, difficult because it would require a lot of cooperation, and consensus which we don't have, and I'm still not sure it would work. But something along those lines might work about as well as web sites that require registration to avoid comment spam - the requirement for "real" user registration blocks comment spam moderately well.
-- Jamie
540kHz is the bottom end of medium wave (MW). Other AM radio sets, especially older or more "international" sets include long wave (LW), which goes down to about 150kHz.
Also, as your assumptions are incorrect, microwave radiation's focus is primarily in a line-of-sight fashion although the omni antennas diffuse this outward to maximize reception in handsets.
:)
:)
;) Only a fraction of the energy of the circular wave will be absorbed even if the boat is a perfect absorber, because as you note most of the wave energy goes elsewhere.
All EM radiation is line of sight for a sufficiently wide beamwidth and sufficiently wide obstacles.
We characterise 30cm microwave radiation as "line-of-sight" for practical transmission because 30cm is small compared with buildings, trees, walls, windows, doors, and other typical obstacles in the environment.
Compare with, say, AM radio which has wavelengths up to 2km at the extreme. Such long waves are barely affected by individual buildings and trees (unless you're inside a conductive building) because of the scale difference - effectively, those waves travel around building-sized obstacles, but are blocked by large structures such as mountains. So we don't describe AM radio as line-of-sight for that reason. (But if you have a space probe and the moon gets in the way, those transmissions are line-of-sight at that scale. It's all relative).
But it is not accurate to think of 30cm microwaves as "line-of-sight" at the scale of a few centimetres - the distance from the cellphone antenna to the LEDs.
A cellphone antenna is omnidirectional largely because you cannot emit a narrow, highly directional beam of 30cm waves from an antenna that fits inside a cellphone.
With current signal processing abilities, if it were possible to emit a highly directional beam (or the equivalent when you take into account multipaths from reflections etc.; that doesn't affect this argument, it just generalises the maths) that would be a huge improvement in spectrum usage, because many phones could use the same frequencies concurrently in the same cell. Just imagine a pencil beam from the base station to each handset - they'd each have a private channel and there would be no interference. But it's not possible because of the physics and the size of handsets relative to the wavelength. It is possible at the base station only, and that's the reason base stations have many antennae pointing in different directions.
It is possible with larger transcievers - and it is done now by MIMO transcievers appearing in current wireless access points - but you'll notice they have at least two antennae, spaced apart (which is essential), and also the wavelength is smaller (about 10cm or 5cm).
(It is actually possible to produce a highly directional beam when several small transcievers cooperate. Through cooperation it is also possible to get a similar increase in data rate for each transciever, as if each one had a directional connection to the base station - a technique I call "distributed lensing". But currently available kit doesn't do that, as far as I know. I'm sure it will, it's the obvious next step after MIMO.)
Back to the point...
On the question of amount of power absorbed: I think we can safely say the power absorbed by the LEDs is, well, the amount of power needed to light a modern blinking LED. Which, as you say, is a few mW at most.
On the analogy with a boat: you need large waves (which you don't get from a throwable rock) from an emitter smaller than the waves (say, a vertical pipe with outlet just below the water surface, pumping water in and out), and a tiny boat that _doesn't_ bob freely; it must be anchored with a tuned, damped spring to make a good energy absorber.
(And if you really meant 359.9 degrees not blocked by the boat, the boat is really far away from the emitter.)
But you make a good point anyway
In the case of large waves from a small transmitter and a good but small absorber close to the transmitter, that is still true: there's a maximum amount which can be absorbed by a perfect absorber, and yes i