Nuclear is not any more dangerous than much of the alternatives...
Tchernobyl and Fukushima is not enough to understand that when nuclear plant go wrong, it's really, really bad for a really, really long time ? Seriously, can you point out other non nuclear energy plant accident that is comparable in term of cost and time to cleanup ? Barrages are probably the most potentially destructive, but even the Banqiao Dam catastrophe have be resolved in less than a generation, unlike the long term isotopes radioactivity of nuclear accident. And the problem do not only cam from damaged nuclear plants, but from every nuclear plan wast.
There not a single example of production size nuclear reactor in the history that have been decommissioned and that you can say that all his wast is now fully harmless to your health. And you will die long long long before someone can claim this. This aspect of the nuclear energy make it very special in term of risk compared to the alternatives.
Switzerland do have industries: http://en.wikipedia.org/wiki/Economy_of_Switzerland#Economic_sectors. I believe that the the biggest advantage of Switzerland is his political stability. Direct democracy, referendum, initiative, and representative government are methods that have helped to get this stability. I hope that more countries will try them.
Your description of the Swiss unemployment insurance is good, but I want to emphasis a few points. Yes there is some discount, but this is really marginal and don't have a big effect on the day to day live. The unemployment insurance last usually for 400 working days, and no, it's really not easy to game this. The control is very effective and the sanction go at high rate. Basically, if you did't found a job the first 3 months, you get much attention from the control and pressure to start formation to make you more in line with the demand. After 6 months, the pressure get even bigger, and the last months you are almost constrain to accept any job out there, even it revenue is fare below what you used to get.
The Swiss unemployment insurance used to be a easy way to get money for lazy peoples 20 years ago back to the day where unemployment was rare, but not anymore. Today the unemployment affect every class of workers, even from banks. The system work because the unemployment rate is still under control here, but the future is actually going into the direction of a higher unemployment rate.
I already have tried that way but i found it not useful.
First the server use a very conservative policy regarding distribution and is because of that constantly hitting the too old barrier about current debootstrap. Second debootstrap only install a very minimal image that require a lot of tweaks until you get all the standard features of a today distribution. Finally debootstrap in not as simple as using the installer. The installer is designed to be simple and it's a good thing.
Proposing a NFS mount is relatively easy. It can even be done automatically if it found the NFS path in the DHCP option and found no disk on the machine.
While it's very easy to use Wheezy as a NFS root filesystem, I found strange that the install still don't allow to mount a NFS share to install a root filesystem on it. I still have to install a real disk on the machine, select it for the installation and then copy his content into the machine's folder on the NFS server. A NFS mount into the installer will make the real disk and the copy no longer required.
The inode information are individual to each file.
Either the purpose is to let system width cleanup regardless of the application that created the files, in witch case inode info is the only practicale soultion because there not grant that every applications will uses the same convention.
Either the purpose is application specific in witch case the responsibility to remove the file belong to the application, not to the system tools.
AF_BUS is completely optional and mostly only a multicast optimization. Nothing that the current D-Bus libraries and daemon can't do for years.
A real advance for D-Bus would be a proper _synchronous_ implementation of the D-Bus into the libglib instead of the actual bloated multithread implementation that can hang in futex. I hit this problem from years in multiple embedded systems. It rare but when it's happen, the product freeze and the customer is unhappy.
No need to add a new date information in the file name. The inode already have the creation, last modification and last reading date. Fare enough to determine witch files are too old.
If you hacked the box, you can certainly also modify the NFS server running on that box.
The NFS server is NOT the same box as the internet server that use NFS root as a client. Here is a simple schema:
RAID array NFS server box Firewall Internet server boxes (NFS root client) Firewall Internet
To archive a permanent modification of one of the internet server box, a file must be modified on the NFS root mounted filesystem. This modification is very easy to detect from the NFS server point of view. I personally use a simple script that parse the rsync log while it backup the internet servers root directories. This setup is very safe as long as the NFS server is not compromised witch require way more advanced hacking if you use a strong firewall between the boxes.
Using NFS root on the internet server make intervention easy: in case of a doubt, just halt the involved internet server. Then you can safely inspect his filesystem root directory on the NFS server, and eventually restore some files before rebooting the internet server. This is very comfortable to do. I add to this that the internet servers are very reliable because there don't have disks on it.
Someone pointed out that the internet server BIOS can still be permanently hacked. If you worry about that, then you can fix the "write protection" signal of the BIOS memory on the motherboard.
I suppose it would be better if somebody came up with a well-developed package for cross-scanning systems over shared storage (does it already exist?) but that's also only going to reduce your 98% to 98.5%. I suppose all you need to really do is to validate the integrity of the host system's scanner (or rpm or dpkg, etc.) but then again the remote system access method could also be compromised (today it's TLA stuff to compromise e.g. nfsd such that it will only return the wrong (but valid) md5sum and sha1sum and [randomhashsum] for only the scanners you might check for) but it's not likely.
Taking down every system on a frequent schedule for an offline scan would be a good idea but then again the BIOS could be compromised and you start to run into the collision of business needs and absolute security.
I use NFS root internet servers connected to a intranet NFS server. On the last I use rsync to do backup and detection of any files change. The only permanent memory on the internet servers machines are the BIOS, ok, but I have yet to see an exploit that hack the BIOS in a way to compromise a TFTP + NFS boot without any detectable trace from the NFS server point of view. If your concern really go into this level of paranoia, then you can simply hack the write protection signal of your BIOS memory chip (see the/WP in of the datasheet) and reboot the internet servers periodically.
I have experimented a very successful way to solve the problem: Internet connected servers using NFS root from an intranet NFS server. All the machines are using a standard distribution. The intranet NFS server handle the backup of all the internet servers using rsync. It's really easy and efficient as it's just moving data internally of the RAID array. It's easy to detect unexpected files change, to compare internet machines binaries with the intranet machine binaries, and to eventually recover them. The downside is that NFS is slow compared to HDD and SSD, but good caching practice should not depend on them anyway. The advantage is that managing all the machines is very easy and comfortable, that the internet server machines are extremely reliable without disks, and that I only have to care about a central RAID array.
The nfs could be hacked to give the original binaries to your outside server.
Absolutely impossible because the hacked file will have his checksum changed from the NFS server point of view. As long as the NFS server is not compromised, this is very safe. I personally use NFS root internet servers since more than 12 years now and I can say that's a very efficient way to detect and recover from external attacks. In addition, it very reliable, comfortable to manage, and easy to backup.
Another key part of the system is the epicycloid gear. On a Prius for example, this gear allow the computer to adjust the energies flows with precision because all is controlled by two electric motors. I wonder how Peugeot Citroen system can archive the same. Possibly by controlling electrovalves on the hydraulic pump and hydraulic motor circuit.
The English article don't show the details. Here is a French article with more details: http://automobile.challenges.fr/dossiers/20130123.LQA4031/psa-hybrid-air-des-hybrides-peugeot-et-citroen-a-air-comprime.html It say that there is two nitrogen tanks, a high pressure one and a low pressure one. A hydraulic fluid (not specified) is used to compress the nitrogen of the high pressure tank. There is no description about the utility of the low pressure tank. There say that the gear boy is actually of the same type of the one used by Toyota in there hybrid cars.
There emphasis the fact that this nitrogen compression can sustain a far higher input energy from the car deceleration compared to the energy that a electric generator can accept. But the total amount of energy stored is 20 less compared to the battery of the Toyota Yaris Hybrid. There say that this amount of energy is not enough to move more than about 100m, but that it can move a lot of power compared to the electric hybrid system.
Look like the goal is to store a lot of energy from deceleration to be able to reuse it for the next acceleration, witch sound good in very urban traffic. I will not be surprised if someone will try in the future to use springs to archive the same goal, like do a lot of toy cars.
Petroleum chemistry don't allow to produce arbitrary ratio of diesel compared to others products. The overall consumption have to match the ration that refineries can produce. So if the the diesel consumption increase, then his price will raise and the others products price will fall. This is exactly what happened in many European countries since the diesel have been promoted. Some government still try to biased the price by the taxes, but this cannot change the petroleum chemistry facts.
Ok, diesel have more couple and a higher stored energy density. But it's not ideal either. Even new cars of European manufacturers that have years of experience with small diesel engine, still make car that are more noisy and smell bad, not counting the small particles that fill the town and are increasingly suspected to play a role in some lung cancer.
Diesel will not disappear, precisely because of the petroleum chemistry, but it's still preferable to use it in heavy engines that can get the required heavy exhaust gas processing without scarify substantial efficiency and without increasing the total cost too much. It's really less comfortable to use a lot of diesel in a town because of the noise, the smell, and the particles.
You should replace "European" in your text by "UK". But UK is an exception about this. All others European countries use SI units: litre and meter. Usually the car consumption is rated in L/100km.
"The pound or pound-mass (abbreviations: lb, lbm, lbm, [1]) is a unit of mass used in the imperial, United States customary and other systems of measurement. A number of different definitions have been used, the most common today being the international avoirdupois pound which is legally defined as exactly 0.45359237 kilograms."
This is not a surprise as the USA have, from the international point of view, endorsed the SI units since a half century.
"The definition of the kilogram is undergoing a fundamental change - the current definition defines the kilogram as being the mass of the international prototype kilogram, the new definition relates it to the equivalent energy of a photon via Planck's constant.
Current definition: The kilogram is the unit of mass; it is equal to the mass of the international prototype of the kilogram.
Proposed definition: The kilogram, kg, is the unit of mass; its magnitude is set by fixing the numerical value of the Planck constant to be equal to exactly 6.62606X×1034 when it is expressed in the unit s1m2kg, which is equal to Js.
One consequence of this change is that the new definition makes the definition of the kilogram dependent on the definitions of the second and the metre."
Slashdot Mirror
Nuclear is not any more dangerous than much of the alternatives ...
Tchernobyl and Fukushima is not enough to understand that when nuclear plant go wrong, it's really, really bad for a really, really long time ? Seriously, can you point out other non nuclear energy plant accident that is comparable in term of cost and time to cleanup ? Barrages are probably the most potentially destructive, but even the Banqiao Dam catastrophe have be resolved in less than a generation, unlike the long term isotopes radioactivity of nuclear accident. And the problem do not only cam from damaged nuclear plants, but from every nuclear plan wast.
There not a single example of production size nuclear reactor in the history that have been decommissioned and that you can say that all his wast is now fully harmless to your health. And you will die long long long before someone can claim this. This aspect of the nuclear energy make it very special in term of risk compared to the alternatives.
Just try to open today a anonymous account in Switzerland, you will be very deceptive. There is others places where it's far more easy:
http://en.wikipedia.org/wiki/Offshore_financial_centre#List_of_main_offshore_financial_centres
Switzerland do have industries: http://en.wikipedia.org/wiki/Economy_of_Switzerland#Economic_sectors. I believe that the the biggest advantage of Switzerland is his political stability. Direct democracy, referendum, initiative, and representative government are methods that have helped to get this stability. I hope that more countries
will try them.
Your description of the Swiss unemployment insurance is good, but I want to emphasis a few points. Yes there is some discount, but this is really marginal and don't have a big effect on the day to day live. The unemployment insurance last usually for 400 working days, and no, it's really not easy to game this. The control is very effective and the sanction go at high rate. Basically, if you did't found a job the first 3 months, you get much attention from the control and pressure to start formation to make you more in line with the demand. After 6 months, the pressure get even bigger, and the last months you are almost constrain to accept any job out there, even it revenue is fare below what you used to get.
The Swiss unemployment insurance used to be a easy way to get money for lazy peoples 20 years ago back to the day where unemployment was rare, but not anymore. Today the unemployment affect every class of workers, even from banks. The system work because the unemployment rate is still under control here, but the future is actually going into the direction of a higher unemployment rate.
I already have tried that way but i found it not useful.
First the server use a very conservative policy regarding distribution and is because of that constantly hitting the too old barrier about current debootstrap. Second debootstrap only install a very minimal image that require a lot of tweaks until you get all the standard features of a today distribution. Finally debootstrap in not as simple as using the installer. The installer is designed to be simple and it's a good thing.
Proposing a NFS mount is relatively easy. It can even be done automatically if it found the NFS path in the DHCP option and found no disk on the machine.
While it's very easy to use Wheezy as a NFS root filesystem, I found strange that the install still don't allow to mount a NFS share to install a root filesystem on it. I still have to install a real disk on the machine, select it for the installation and then copy his content into the machine's folder on the NFS server. A NFS mount into the installer will make the real disk and the copy no longer required.
The inode information are individual to each file.
Either the purpose is to let system width cleanup regardless of the application that created the files, in witch case inode info is the only practicale soultion because there not grant that every applications will uses the same convention.
Either the purpose is application specific in witch case the responsibility to remove the file belong to the application, not to the system tools.
AF_BUS is completely optional and mostly only a multicast optimization. Nothing that the current D-Bus libraries and daemon can't do for years.
A real advance for D-Bus would be a proper _synchronous_ implementation of the D-Bus into the libglib instead of the actual bloated multithread implementation that can hang in futex. I hit this problem from years in multiple embedded systems. It rare but when it's happen, the product freeze and the customer is unhappy.
Mod parent up.
No need to add a new date information in the file name. The inode already have the creation, last modification and last reading date. Fare enough to determine witch files are too old.
Arrg! Missed to see that the formatting removed chars on the graph. Below is a correct one:
RAID array ==== NFS server box ==== Firewall ==== Internet server boxes (NFS root client) ==== Firewall ==== Internet
If you hacked the box, you can certainly also modify the NFS server running on that box.
The NFS server is NOT the same box as the internet server that use NFS root as a client. Here is a simple schema:
RAID array NFS server box Firewall Internet server boxes (NFS root client) Firewall Internet
To archive a permanent modification of one of the internet server box, a file must be modified on the NFS root mounted filesystem. This modification is very easy to detect from the NFS server point of view. I personally use a simple script that parse the rsync log while it backup the internet servers root directories. This setup is very safe as long as the NFS server is not compromised witch require way more advanced hacking if you use a strong firewall between the boxes.
Using NFS root on the internet server make intervention easy: in case of a doubt, just halt the involved internet server. Then you can safely inspect his filesystem root directory on the NFS server, and eventually restore some files before rebooting the internet server. This is very comfortable to do. I add to this that the internet servers are very reliable because there don't have disks on it.
Someone pointed out that the internet server BIOS can still be permanently hacked. If you worry about that, then you can fix the "write protection" signal of the BIOS memory on the motherboard.
I suppose it would be better if somebody came up with a well-developed package for cross-scanning systems over shared storage (does it already exist?) but that's also only going to reduce your 98% to 98.5%. I suppose all you need to really do is to validate the integrity of the host system's scanner (or rpm or dpkg, etc.) but then again the remote system access method could also be compromised (today it's TLA stuff to compromise e.g. nfsd such that it will only return the wrong (but valid) md5sum and sha1sum and [randomhashsum] for only the scanners you might check for) but it's not likely.
Taking down every system on a frequent schedule for an offline scan would be a good idea but then again the BIOS could be compromised and you start to run into the collision of business needs and absolute security.
I use NFS root internet servers connected to a intranet NFS server. On the last I use rsync to do backup and detection of any files change. The only permanent memory on the internet servers machines are the BIOS, ok, but I have yet to see an exploit that hack the BIOS in a way to compromise a TFTP + NFS boot without any detectable trace from the NFS server point of view. If your concern really go into this level of paranoia, then you can simply hack the write protection signal of your BIOS memory chip (see the /WP in of the datasheet) and reboot the internet servers periodically.
Yes, "out-of-tree" is the key.
I have experimented a very successful way to solve the problem: Internet connected servers using NFS root from an intranet NFS server. All the machines are using a standard distribution. The intranet NFS server handle the backup of all the internet servers using rsync. It's really easy and efficient as it's just moving data internally of the RAID array. It's easy to detect unexpected files change, to compare internet machines binaries with the intranet machine binaries, and to eventually recover them. The downside is that NFS is slow compared to HDD and SSD, but good caching practice should not depend on them anyway. The advantage is that managing all the machines is very easy and comfortable, that the internet server machines are extremely reliable without disks, and that I only have to care about a central RAID array.
Simply use NFS root for the internet server. It's reliable, easy to backup, and fun to manage.
The nfs could be hacked to give the original binaries to your outside server.
Absolutely impossible because the hacked file will have his checksum changed from the NFS server point of view. As long as the NFS server is not compromised, this is very safe. I personally use NFS root internet servers since more than 12 years now and I can say that's a very efficient way to detect and recover from external attacks. In addition, it very reliable, comfortable to manage, and easy to backup.
Thanks for all the details.
Another key part of the system is the epicycloid gear. On a Prius for example, this gear allow the computer to adjust the energies flows with precision because all is controlled by two electric motors. I wonder how Peugeot Citroen system can archive the same. Possibly by controlling electrovalves on the hydraulic pump and hydraulic motor circuit.
Really sorry for my English. You are right it's not my native language.
My native language is French.
Also, the signs on the motorway that say "2/3 mile" are placed exactly 1km from the exit, they are just labelled wrong.
This is certainly a clever decision in case the UK roads will switch to the km notation in a possible future.
The English article don't show the details. Here is a French article with more details:
http://automobile.challenges.fr/dossiers/20130123.LQA4031/psa-hybrid-air-des-hybrides-peugeot-et-citroen-a-air-comprime.html
It say that there is two nitrogen tanks, a high pressure one and a low pressure one. A hydraulic fluid (not specified) is used to compress the nitrogen of the high pressure tank. There is no description about the utility of the low pressure tank. There say that the gear boy is actually of the same type of the one used by Toyota in there hybrid cars.
There emphasis the fact that this nitrogen compression can sustain a far higher input energy from the car deceleration compared to the energy that a electric generator can accept. But the total amount of energy stored is 20 less compared to the battery of the Toyota Yaris Hybrid. There say that this amount of energy is not enough to move more than about 100m, but that it can move a lot of power compared to the electric hybrid system.
Look like the goal is to store a lot of energy from deceleration to be able to reuse it for the next acceleration, witch sound good in very urban traffic. I will not be surprised if someone will try in the future to use springs to archive the same goal, like do a lot of toy cars.
Petroleum chemistry don't allow to produce arbitrary ratio of diesel compared to others products. The overall consumption have to match the ration that refineries can produce. So if the the diesel consumption increase, then his price will raise and the others products price will fall. This is exactly what happened in many European countries since the diesel have been promoted. Some government still try to biased the price by the taxes, but this cannot change the petroleum chemistry facts.
Ok, diesel have more couple and a higher stored energy density. But it's not ideal either. Even new cars of European manufacturers that have years of experience with small diesel engine, still make car that are more noisy and smell bad, not counting the small particles that fill the town and are increasingly suspected to play a role in some lung cancer.
Diesel will not disappear, precisely because of the petroleum chemistry, but it's still preferable to use it in heavy engines that can get the required heavy exhaust gas processing without scarify substantial efficiency and without increasing the total cost too much. It's really less comfortable to use a lot of diesel in a town because of the noise, the smell, and the particles.
You should replace "European" in your text by "UK". But UK is an exception about this.
All others European countries use SI units: litre and meter. Usually the car consumption is rated in L/100km.
http://en.wikipedia.org/wiki/New_SI_definitions#Kilogram
It seem that the legal definition of pound is bound to the kilogram: http://en.wikipedia.org/wiki/Pound_(mass)
"The pound or pound-mass (abbreviations: lb, lbm, lbm, [1]) is a unit of mass used in the imperial, United States customary and other systems of measurement. A number of different definitions have been used, the most common today being the international avoirdupois pound which is legally defined as exactly 0.45359237 kilograms."
This is not a surprise as the USA have, from the international point of view, endorsed the SI units since a half century.
http://en.wikipedia.org/wiki/New_SI_definitions#Kilogram
http://en.wikipedia.org/wiki/New_SI_definitions
"The definition of the kilogram is undergoing a fundamental change - the current definition defines the kilogram as being the mass of the international prototype kilogram, the new definition relates it to the equivalent energy of a photon via Planck's constant.
Current definition: The kilogram is the unit of mass; it is equal to the mass of the international prototype of the kilogram.
Proposed definition: The kilogram, kg, is the unit of mass; its magnitude is set by fixing the numerical value of the Planck constant to be equal to exactly 6.62606X×1034 when it is expressed in the unit s1m2kg, which is equal to Js.
One consequence of this change is that the new definition makes the definition of the kilogram dependent on the definitions of the second and the metre."
Thanks for the explanation, I misinterpreted the 'gun shoot' words.