Do you want to have a secure job maintaining those legacy systems like maintaining IE6 only webapps? Then fine go for it. If you want to have something that should still be used easily in 10 years, by all means no.
If you want something serious, use some cross-platform development system like Java or Lazarus and compile for whatever you want. Or you make a web application with a more abstract interface. For example by separating the user interface from the application logic.
Metro simply is yet another vendor lock-in in a world that has moved on years ago. Even VT100 terminals have a brighter future than it.
Particularly considering it's just a "simple" tax programme. It sounds like, instead of implementing the logic in data (i.e. arrays, external text files, etc) and having a fairly small compiler targeted at the problem, they implemented everything in C. This is something even C advocates advise against.
Well of course, but a transparent calculation can be checked and it adds at least a little bit more credibility than pure guessing. If you have one calculation supporting one opinion and one supporting the other you can compare both and look for the differences.
Something which is suspiciously missing from the whole discussion are facts. I mean if you claim that nuclear power emits less CO2 per kWh, it should be trivial to back that up with facts. You would simply add up all the CO2 emitted by the mining of the fuels to the CO2 emitted during the construction of the plant, and the CO2 emitted by the cars of the workers there up to the CO2 caused by the disposal of the waste, as well as all those little things I just missed.
If you add up all those points and also list them, only then you can make such an argument.
Actually you can type editor on most unixoid systems, and get to your pre-set editor.
If you mistype something, you'll get a prompt "command not found, did you mean perhaps x?". If you press tab twice you'll get a list of all options, etc.
With a decent command line you can simply do all the things people do on Excel databases in simple text files and shell scripts. And since that's all open source and uses open formats, changing to something better is trivial. Or you can simply add a different frontend.
If you have less than 100k datasets in your database, linear search is still acceptable.
My hypothesis is that those people previously worked inside the Windows ecosystem. There complexity is a treasured asset. The more complex a system is the better it seems to be. That's the Windows philosophy.
The Unix philosophy is different. Everything should be as simple and elegant as possible. Data should be stored in text format, whenever possible.
If you want to see a typical Windows way of solving a problem. Look at OPC. The OLE for Process Control. Essentially that's a way to communicate with an object storage system. Networking is done via DCOM. Even trivial things become complex.
Because a command line allows you to work way more efficient than any GUI could ever do. It simply allows you to give your system commands of any complexity. This can be anything from a simple command, to something very complex.
Maybe one day someone will come up with a usable way of doing that via a GUI, but until then there simply is no alternative for serious users.
(Of course there are now a lot of people who are afraid of computers and just use them as word processors, video recorders, gaming machines or other tasks more suited to dedicated machines.
There is no reason why running a server should have disadvantages for your ISP. If there was competition among ISPs, no ISP would impose such a restriction.
Other than that, a small virtual host at Funkfeuer.at (Austrian ISP made to finance their Freifunk project) only costs 9 Euros a month. It's even cheaper at commercial providers.
I'm also considering this as a possibility because I might have to move next year and it would be nice for the mailserver to stay up then.
How does that work? Do you put the vote into an urn? How do you know the voting computer counted the vote the way it was printed?
If you put the vote into an urn, why don't you just replace the voting computer with a pen? Germany proves that to be very efficient with most voting locales finishing their duty 30 minutes after the end of the election.
Building a voting computer which satisfies the demands for a democratic election is near impossible. Since fraud needs to be detectable even by single uneducated voters, there minimum security would be like this:
1. Get at least 80% of your voters a degree in Mathematics and Cryptology. They need to be able to verify all the algorithms used in the process. 2. Get at least 80% of your voters fluent in reading machine code off microscope images of ROM chips. 3. Get at least 80% of your voters good at re-engineering micro controller systems from silicon up in a reasonable timespan. (e.g. 30 minutes, this might require genetic engineering) 4. Develop a form of computing device which is transparent.
The big point is, it's not enough if we have some "perfect" voting computer which 10 specialists attest to be "perfect". For a democratic election everybody who is allowed to vote must be able to check the system for fraud. With a simple pen and paper system that is trivial. You just sit at the polling station, check that only single sheets are handed out to the voters. You also check that the voting urn is empty when the voting starts and that everybody just puts in his single sheet into it. Then you check the counting for miscounts and people trying to hide votes. The total number of votes can be compared in different ways.
So everybody involved in it can check it. There is no secret knownledge involved. You can come up with the points I just wrote by yourself. You can even find the points I was missing. That's the minimum standard for voting systems, and it can be settled by the cheapest way to conduct elections, pen and paper. Why on earth should we spend a lot of money for much worse systems?
If you have a high budget, rent a flat in the UK, where you can set up a satellite dish. Then use VDR and a high speed (>10 MBit) Internet connection to where you live to get to the signal.
Other than that, certainly get some video capable computer connected to your TV. If you are in the US, try to get free to air satellite. Unfortunately because of the stupidity of the US market, you are pretty much out of luck when it comes to satellite television. Maybe you can erect some tower in order to get some terrestrial channels and use MythTV.
Well the ROM BIOS is executed by the main CPU. Of course if you want to have binary firmware on the PCI-Bus you need to have a way to restrict the access rights of the hardware. Again that's a lot harder than simply releasing the specifications for the hardware.
Well the binary blobs would then be only firmware, but nothing which runs on your main CPU. This solves a lot of the problems. For example you get a security barrier between the proprietary code and your kernel.
I mean the obvious solution would be open source code, but nVidia doesn't want that.
You misunderstood. The OpenGL support should move from the driver to the hardware so you won't need a proprietary driver any more. Essentially every graphics card would have (more or less) the same interface to the operating system. No hardware dependent drivers would be needed then.
Put a microprocessor on the card which accepts OpenGL commands (or something similar) from the computer and does whatever proprietary stuff you would usually do on the card. Then publish and open the interface between the host system and that controller, so every graphics card will look "the same" to the host system. Get the complexity out of the driver and into the "hardware".
The whole idea of closed source software is ridiculous anyhow. Don't you think your competitors haven't already analysed it in detail just like they reverse engineered your chips for years?
A password you cannot change but can read out 6 meters away without the victim knowing, isn't a good idea. Fingerprints are useless for security and this project simply proves that. You can do, at most, identification, but not authentication.
Electronic Voting cannot be democratic as it doesn't conform to the minimal standards.
So far nobody has proposed an electronic voting system which can be proven to not be manipulated by anybody. If you need a degree in math to understand how the security works, it may be suitable for an election in the maths department of an university, but it is not suitable for the general population.
The pen an paper system can be checked by everybody, not just specialists who might fear for their job if they became politically active.
Actually patent attorneys are fairly useless. What you need to do is to follow the procedures accurately, even down to stupid things like font sizes and such. Then when it gets rejected, you address the points made in the rejection and re-submit it. It usually simply passes then. If not, the examiner will have raised valid points you need to address again.
Note that patents usually are worthless. Unless they are illegally broad, it's trivial to work around them. If they are not worked around, they typically are useless. The patents that do generate money are the ones referenced in standards or the ones so trivial people stumble across them.
Well since universal lossless compression could even compress its output, (if it existed) you can suddenly store infinite amounts of data on finite amounts of space for a finite budget. There are many institutions which have to store a lot of data. They could get rid of racks of tapes or harddisks and just use some tiny little storage device for their archive.
However as pointed out before, something like that is logically impossible. That however doesn't keep stupid people from trying anyhow. The current crop of algorithms like bzip2 are already so close to the maximum possible that putting in more effort is futile.
Slashdot Mirror
Do you want to have a secure job maintaining those legacy systems like maintaining IE6 only webapps? Then fine go for it. If you want to have something that should still be used easily in 10 years, by all means no.
If you want something serious, use some cross-platform development system like Java or Lazarus and compile for whatever you want. Or you make a web application with a more abstract interface. For example by separating the user interface from the application logic.
Metro simply is yet another vendor lock-in in a world that has moved on years ago. Even VT100 terminals have a brighter future than it.
Particularly considering it's just a "simple" tax programme.
It sounds like, instead of implementing the logic in data (i.e. arrays, external text files, etc) and having a fairly small compiler targeted at the problem, they implemented everything in C. This is something even C advocates advise against.
Just measure how much code lines you have spent. The less code a project needed, giving a certain set of features, the better it is.
Well of course, but a transparent calculation can be checked and it adds at least a little bit more credibility than pure guessing. If you have one calculation supporting one opinion and one supporting the other you can compare both and look for the differences.
Something which is suspiciously missing from the whole discussion are facts. I mean if you claim that nuclear power emits less CO2 per kWh, it should be trivial to back that up with facts. You would simply add up all the CO2 emitted by the mining of the fuels to the CO2 emitted during the construction of the plant, and the CO2 emitted by the cars of the workers there up to the CO2 caused by the disposal of the waste, as well as all those little things I just missed.
If you add up all those points and also list them, only then you can make such an argument.
that virtually all the found backdoors originated in the US. Virtually no Chinese back doors have been found yet.
Well OSX isn't really an alternative to Windows. It's done by people with nearly the same mindset.
Actually you can type editor on most unixoid systems, and get to your pre-set editor.
If you mistype something, you'll get a prompt "command not found, did you mean perhaps x?".
If you press tab twice you'll get a list of all options, etc.
With a decent command line you can simply do all the things people do on Excel databases in simple text files and shell scripts. And since that's all open source and uses open formats, changing to something better is trivial. Or you can simply add a different frontend.
If you have less than 100k datasets in your database, linear search is still acceptable.
My hypothesis is that those people previously worked inside the Windows ecosystem. There complexity is a treasured asset. The more complex a system is the better it seems to be. That's the Windows philosophy.
The Unix philosophy is different. Everything should be as simple and elegant as possible. Data should be stored in text format, whenever possible.
If you want to see a typical Windows way of solving a problem. Look at OPC. The OLE for Process Control. Essentially that's a way to communicate with an object storage system. Networking is done via DCOM. Even trivial things become complex.
Because a command line allows you to work way more efficient than any GUI could ever do.
It simply allows you to give your system commands of any complexity. This can be anything from a simple command, to something very complex.
Maybe one day someone will come up with a usable way of doing that via a GUI, but until then there simply is no alternative for serious users.
(Of course there are now a lot of people who are afraid of computers and just use them as word processors, video recorders, gaming machines or other tasks more suited to dedicated machines.
There is no reason why running a server should have disadvantages for your ISP. If there was competition among ISPs, no ISP would impose such a restriction.
Other than that, a small virtual host at Funkfeuer.at (Austrian ISP made to finance their Freifunk project) only costs 9 Euros a month. It's even cheaper at commercial providers.
I'm also considering this as a possibility because I might have to move next year and it would be nice for the mailserver to stay up then.
How does that work? Do you put the vote into an urn? How do you know the voting computer counted the vote the way it was printed?
If you put the vote into an urn, why don't you just replace the voting computer with a pen? Germany proves that to be very efficient with most voting locales finishing their duty 30 minutes after the end of the election.
Building a voting computer which satisfies the demands for a democratic election is near impossible.
Since fraud needs to be detectable even by single uneducated voters, there minimum security would be like this:
1. Get at least 80% of your voters a degree in Mathematics and Cryptology. They need to be able to verify all the algorithms used in the process.
2. Get at least 80% of your voters fluent in reading machine code off microscope images of ROM chips.
3. Get at least 80% of your voters good at re-engineering micro controller systems from silicon up in a reasonable timespan. (e.g. 30 minutes, this might require genetic engineering)
4. Develop a form of computing device which is transparent.
The big point is, it's not enough if we have some "perfect" voting computer which 10 specialists attest to be "perfect". For a democratic election everybody who is allowed to vote must be able to check the system for fraud. With a simple pen and paper system that is trivial. You just sit at the polling station, check that only single sheets are handed out to the voters. You also check that the voting urn is empty when the voting starts and that everybody just puts in his single sheet into it. Then you check the counting for miscounts and people trying to hide votes. The total number of votes can be compared in different ways.
So everybody involved in it can check it. There is no secret knownledge involved. You can come up with the points I just wrote by yourself. You can even find the points I was missing. That's the minimum standard for voting systems, and it can be settled by the cheapest way to conduct elections, pen and paper. Why on earth should we spend a lot of money for much worse systems?
If you have a high budget, rent a flat in the UK, where you can set up a satellite dish. Then use VDR and a high speed (>10 MBit) Internet connection to where you live to get to the signal.
Other than that, certainly get some video capable computer connected to your TV. If you are in the US, try to get free to air satellite. Unfortunately because of the stupidity of the US market, you are pretty much out of luck when it comes to satellite television. Maybe you can erect some tower in order to get some terrestrial channels and use MythTV.
Well the ROM BIOS is executed by the main CPU. Of course if you want to have binary firmware on the PCI-Bus you need to have a way to restrict the access rights of the hardware. Again that's a lot harder than simply releasing the specifications for the hardware.
Well the binary blobs would then be only firmware, but nothing which runs on your main CPU. This solves a lot of the problems. For example you get a security barrier between the proprietary code and your kernel.
I mean the obvious solution would be open source code, but nVidia doesn't want that.
You misunderstood. The OpenGL support should move from the driver to the hardware so you won't need a proprietary driver any more. Essentially every graphics card would have (more or less) the same interface to the operating system. No hardware dependent drivers would be needed then.
Put a microprocessor on the card which accepts OpenGL commands (or something similar) from the computer and does whatever proprietary stuff you would usually do on the card. Then publish and open the interface between the host system and that controller, so every graphics card will look "the same" to the host system. Get the complexity out of the driver and into the "hardware".
The whole idea of closed source software is ridiculous anyhow. Don't you think your competitors haven't already analysed it in detail just like they reverse engineered your chips for years?
A password you cannot change but can read out 6 meters away without the victim knowing, isn't a good idea. Fingerprints are useless for security and this project simply proves that. You can do, at most, identification, but not authentication.
Electronic Voting cannot be democratic as it doesn't conform to the minimal standards.
So far nobody has proposed an electronic voting system which can be proven to not be manipulated by anybody. If you need a degree in math to understand how the security works, it may be suitable for an election in the maths department of an university, but it is not suitable for the general population.
The pen an paper system can be checked by everybody, not just specialists who might fear for their job if they became politically active.
Yes, like I said it's sad. If you can speak German you can look up "Kryptochef".
http://kryptochef.net/indexh2e.htm
Man it would be great if someone could do some "krypto" analysis on his product. That would be a fun story.
Well there are people who claim that they can compress _everything_ by a factor of X. And those are the people we are talking about.
Actually patent attorneys are fairly useless.
What you need to do is to follow the procedures accurately, even down to stupid things like font sizes and such. Then when it gets rejected, you address the points made in the rejection and re-submit it. It usually simply passes then. If not, the examiner will have raised valid points you need to address again.
Note that patents usually are worthless. Unless they are illegally broad, it's trivial to work around them. If they are not worked around, they typically are useless. The patents that do generate money are the ones referenced in standards or the ones so trivial people stumble across them.
Well since universal lossless compression could even compress its output, (if it existed) you can suddenly store infinite amounts of data on finite amounts of space for a finite budget. There are many institutions which have to store a lot of data. They could get rid of racks of tapes or harddisks and just use some tiny little storage device for their archive.
However as pointed out before, something like that is logically impossible. That however doesn't keep stupid people from trying anyhow.
The current crop of algorithms like bzip2 are already so close to the maximum possible that putting in more effort is futile.