In a month or two(was in print several months ago), Linux Journal is going to have an article on how to create a USB device that communicates through the Linux USB generic serial driver.
I know it doesn't solve everyone's problems, but you're free to submit a patch to allow the same functionality from the normal serial device to the USB one.
If the tax is per-gigabit, then the circuit provider is screwed. They can't measure traffix over their line. Only the service provider can.
The circuit provider would have to get the count from the service provider. Considering that the service provider isn't the one being taxed, the circuit provider is going to charge per-gigabit rates.
That brings up a legality issue. Is it legal to charge a client on a per-unit basis where the unit isn't even what the provider's service is measured in?
The idea of selling more resources than you have comes from dial-up ISPs, where you'd have a lot more customers than you had phone lines. Most of your custumers won't need to be connected at the same time.
Cable and DSL customers followed dial-up's cost-saving policy but it's now biting them in the arse.
When always-on connections came around, people started running always-on applications.
It might be interesting to think about the cost back when the PSTN was a switched circuit network. You weren't so much limited by data throughput as you were by the number of circuits that are available.
Bye the way, "PSTN" stands for "Public Switched Telephone Network."
The whole argument is that America is becoming a more service-oriented industry, rather than a product-oriented industry. Sales taxes are grabbing less and less of the total economy.
Sales tax worked well in the 50s, when all your needs were purchased goods.
Also, if you raise taxes on physical goods, you end up getting more tax money from the people who depend more on physical goods than on services. Everyone needs clothes, not everyone needs a broadband internet connection.
Selling more total resources than you have is common in the ISP business. The ISP is betting that their customers won't all want to peak their bandwidth usage at the same time.
Now, it is definitely possible to oversell your resources. Take a look at AOL, for example.
That doesn't work, since one company normally provides the line, while several different companies attempt to provide the service. The company providing the line would get really pissed, since they're not the ones that have any control over usage.
Just because my T1 circuit goes through SBC doesn't mean I'm not getting the service itself from another company.
The government should take notice of all the warning signs. But then, only the FBI, CIA and NSA are given budgets for that task.
This is the first time I've heard of an "independant panel of experts" have something negative to say about dependance on Microsoft.
And it gives me an idea...The next time there is a "cyber-terrorism" scare, write your national congressmen and senators about the dangers to National Security posed by dependence on Microsoft.
Then follow up with a suggestion recommending open-source software.
And be sure to document the fact. Evidence of influence by Microsoft during a vote, in such times, could be rather damning next election.
(It's harder for virus makers to affect more computers at once if less computers use the same OS)
Even in one OS, if it's an OS where swapping service-providing programs is common, the same exploit won't necessarily be present on two different machines.
For example, you can replace Sendmail with Exim, or Mozilla with Ximian-Evolution.
Hell, under Windows, you can even replace IIS with Apache.:)
The funny thing about using for loops is that gcc can optimise the loop right out of existance. Assuming you don't optimize your code, you'll still have to deal with branch prediction in the processor.
I don't see anyone thinking about the potential for abuse, with distributed RBLs.
First, if anonymous submissions are allowed, then anyone with a petty vendetta against someone (read individual or network), then they can add them to the RBL.
Now, I can see modifying Gnutella clients to keep a local database of RBL entries and who submitted them in an SQL database. Then you'd periodically generate the local RBL list from an SQL query with an "allow from host" or "deny from host" list.
Ah, the beauty of Open Source. I'd really like to see that happen.
The "black boxes" in airplanes are danger-yellow with black stripes. Would a judge rule on its used based on the wording ("black box") or on the intent? ("status recording device")
(I've never seen the actual law, they might be more specific than "black box".)
Over long periods of time, the assumed intent of a law can be questioned. (Think "right to bear arms") Also, over long periods of time, the wording of a law may not mean the same thing.
Oh, and a hearty "hear hear" on the need for driver's ed improvement.
Look at all the training you have to go through to get a pilot's license. I don't think you could do much more damage flying a single engine prop plane recklessly for an hour per week than driving a corvette recklessly through the suburbs for two hours a day.
Conversly, black boxes could be used to prove your innosence. Cop says you were going 5 over? Pull up your handly Linux interface terminal and say, "See officer? I've been going 55 for the past four miles, except where I passed a guy there, and where a guy braked to turn in front of me."
I've been thinking about installing such a device in my car, just for that reason.
My real concern in a terrorism investigation lay in federal "investigators" confiscating your black box as evidence, when that "evidence" would clear your name. I wouldn't put it past them.
Re:Microsoft prices its operating systems per CPU
on
Is Prescott 64-bit?
·
· Score: 1
That's a good question. And we're seeing some of the results.
Best Buy and friends are selling machines with XP Home installed. These don't take advantage of HyperThreading on the latest Pentium 4s. People who are aware of the fact go out and buy XP Pro, but the rest don't know.
PC Vendors can probably continue to get away with selling "crippled" PCs.
An interesting technical question lays in whether the OS can control applications' access to multibundle functionality in the processor? I can only see that as possible if Microsoft comes out with another executable format, where bundle execution is scheduled by the OS.
Alternatively, they could require that everything be run under the.NET interpereter, and control system usage from there.
Sure, dependencies can be an issue. But saying that upgrading and patching *nix platforms does not produce any mind-numbing dependency issues is simply a self delusion.
I used to run Debian Woody (stable). At the cost of not getting the latest features, I got all the stability I could ask for, with all of the security patches backported. Now I run Debian Sid (unstable), which tends to have dependency problems, but I at least have control over them. And if worse comes to worst, I can install the patches personally.
Open Source has less SEVERE security issues, because of how many patches are made available. Issues in Open Source software tend to be patched quicker than in closed software.
Open Source and UNIX-like environments are an even better combination. With Microsoft, it seems as though every time you patch, something else gets broken. (This is especially true with service packs to Windows and Visual Studio.) With UNIX environments and their separation of tasks into different programs, patching OpenSSH isn't going to bork XFree86.
Re:Microsoft prices its operating systems per CPU
on
Is Prescott 64-bit?
·
· Score: 1
My point is that multibundle CPUs, multilogical CPUs, and multiprocessor systems lend themselves to server tasks.
Windows is priced on the theory that, as far as a customer is concerned, a parallel-processing system is going to be worth paying extra for the OS.
With that in mind, I don't think Microsoft will want to soften its pricing system for a new CPU architecture. Assuming there's a way to detect how many simultaneous bundles an IA-64 CPU can handle, I see Microsoft preferring to charge different amounts for an OS that can support a different level processor.
I've run Robot Odyssey for DOS on my Duron 750MHz.
Short review? Don't bother. In the innovation lab, your batteries die in seconds.
The moral here is that software hasn't always been forwards compatible. It used to be very noticable in timing routines, running Dune 2 on a fast 386, or, god forbid, a 486.
(For the Robot Odyssey example, play DroidQuest instead. It's a Java port of RO.)
I expect to see a lot of similar problems with OSS software. Not because of timing routines, but because of unclean code. For those who are accustomed to writing 32-bit code, making and keeping that code 64-bit clean is going to be a concious effort.
Slashdot Mirror
If you tax per line, you're right. If you tax per bandwidth used, as the comment suggests, then you run into problems.
In a month or two(was in print several months ago), Linux Journal is going to have an article on how to create a USB device that communicates through the Linux USB generic serial driver.
I know it doesn't solve everyone's problems, but you're free to submit a patch to allow the same functionality from the normal serial device to the USB one.
Wait till they're burned out before you change them. Or turn off the light switch first.
If the tax is per-gigabit, then the circuit provider is screwed. They can't measure traffix over their line. Only the service provider can.
The circuit provider would have to get the count from the service provider. Considering that the service provider isn't the one being taxed, the circuit provider is going to charge per-gigabit rates.
That brings up a legality issue. Is it legal to charge a client on a per-unit basis where the unit isn't even what the provider's service is measured in?
You definitely have a point.
The idea of selling more resources than you have comes from dial-up ISPs, where you'd have a lot more customers than you had phone lines. Most of your custumers won't need to be connected at the same time.
Cable and DSL customers followed dial-up's cost-saving policy but it's now biting them in the arse.
When always-on connections came around, people started running always-on applications.
Part of managing an economy lies in encouraging some industries and discouraging others.
Didn't you ever play SimCity 2000?
It might be interesting to think about the cost back when the PSTN was a switched circuit network. You weren't so much limited by data throughput as you were by the number of circuits that are available.
Bye the way, "PSTN" stands for "Public Switched Telephone Network."
The whole argument is that America is becoming a more service-oriented industry, rather than a product-oriented industry. Sales taxes are grabbing less and less of the total economy.
Sales tax worked well in the 50s, when all your needs were purchased goods.
Also, if you raise taxes on physical goods, you end up getting more tax money from the people who depend more on physical goods than on services. Everyone needs clothes, not everyone needs a broadband internet connection.
Selling more total resources than you have is common in the ISP business. The ISP is betting that their customers won't all want to peak their bandwidth usage at the same time.
Now, it is definitely possible to oversell your resources. Take a look at AOL, for example.
That doesn't work, since one company normally provides the line, while several different companies attempt to provide the service. The company providing the line would get really pissed, since they're not the ones that have any control over usage.
Just because my T1 circuit goes through SBC doesn't mean I'm not getting the service itself from another company.
The USS Yorktown had to be towed to port due to NT crashing. I can't find the original news articles, though.
The government should take notice of all the warning signs. But then, only the FBI, CIA and NSA are given budgets for that task.
This is the first time I've heard of an "independant panel of experts" have something negative to say about dependance on Microsoft.
And it gives me an idea...The next time there is a "cyber-terrorism" scare, write your national congressmen and senators about the dangers to National Security posed by dependence on Microsoft.
Then follow up with a suggestion recommending open-source software.
And be sure to document the fact. Evidence of influence by Microsoft during a vote, in such times, could be rather damning next election.
(It's harder for virus makers to affect more computers at once if less computers use the same OS)
:)
Even in one OS, if it's an OS where swapping service-providing programs is common, the same exploit won't necessarily be present on two different machines.
For example, you can replace Sendmail with Exim, or Mozilla with Ximian-Evolution.
Hell, under Windows, you can even replace IIS with Apache.
...what those high-voltage inverters for the cold cathode flourescents do to the delicate machines they're in.
The funny thing about using for loops is that gcc can optimise the loop right out of existance. Assuming you don't optimize your code, you'll still have to deal with branch prediction in the processor.
I don't see anyone thinking about the potential for abuse, with distributed RBLs.
First, if anonymous submissions are allowed, then anyone with a petty vendetta against someone (read individual or network), then they can add them to the RBL.
Now, I can see modifying Gnutella clients to keep a local database of RBL entries and who submitted them in an SQL database. Then you'd periodically generate the local RBL list from an SQL query with an "allow from host" or "deny from host" list.
Ah, the beauty of Open Source. I'd really like to see that happen.
The "black boxes" in airplanes are danger-yellow with black stripes. Would a judge rule on its used based on the wording ("black box") or on the intent? ("status recording device")
(I've never seen the actual law, they might be more specific than "black box".)
Over long periods of time, the assumed intent of a law can be questioned. (Think "right to bear arms") Also, over long periods of time, the wording of a law may not mean the same thing.
A black box in my car may not obtrusive as satellite tracking or a camera on every street corner, but it's still the same principle.
Oh, and a hearty "hear hear" on the need for driver's ed improvement.
Look at all the training you have to go through to get a pilot's license. I don't think you could do much more damage flying a single engine prop plane recklessly for an hour per week than driving a corvette recklessly through the suburbs for two hours a day.
Conversly, black boxes could be used to prove your innosence. Cop says you were going 5 over? Pull up your handly Linux interface terminal and say, "See officer? I've been going 55 for the past four miles, except where I passed a guy there, and where a guy braked to turn in front of me."
I've been thinking about installing such a device in my car, just for that reason.
My real concern in a terrorism investigation lay in federal "investigators" confiscating your black box as evidence, when that "evidence" would clear your name. I wouldn't put it past them.
That's a good question. And we're seeing some of the results.
.NET interpereter, and control system usage from there.
Best Buy and friends are selling machines with XP Home installed. These don't take advantage of HyperThreading on the latest Pentium 4s. People who are aware of the fact go out and buy XP Pro, but the rest don't know.
PC Vendors can probably continue to get away with selling "crippled" PCs.
An interesting technical question lays in whether the OS can control applications' access to multibundle functionality in the processor? I can only see that as possible if Microsoft comes out with another executable format, where bundle execution is scheduled by the OS.
Alternatively, they could require that everything be run under the
Sure, dependencies can be an issue. But saying that upgrading and patching *nix platforms does not produce any mind-numbing dependency issues is simply a self delusion.
I used to run Debian Woody (stable). At the cost of not getting the latest features, I got all the stability I could ask for, with all of the security patches backported. Now I run Debian Sid (unstable), which tends to have dependency problems, but I at least have control over them. And if worse comes to worst, I can install the patches personally.
Open Source has less SEVERE security issues, because of how many patches are made available. Issues in Open Source software tend to be patched quicker than in closed software.
Open Source and UNIX-like environments are an even better combination. With Microsoft, it seems as though every time you patch, something else gets broken. (This is especially true with service packs to Windows and Visual Studio.) With UNIX environments and their separation of tasks into different programs, patching OpenSSH isn't going to bork XFree86.
My point is that multibundle CPUs, multilogical CPUs, and multiprocessor systems lend themselves to server tasks.
Windows is priced on the theory that, as far as a customer is concerned, a parallel-processing system is going to be worth paying extra for the OS.
With that in mind, I don't think Microsoft will want to soften its pricing system for a new CPU architecture. Assuming there's a way to detect how many simultaneous bundles an IA-64 CPU can handle, I see Microsoft preferring to charge different amounts for an OS that can support a different level processor.
I've run Robot Odyssey for DOS on my Duron 750MHz.
Short review? Don't bother. In the innovation lab, your batteries die in seconds.
The moral here is that software hasn't always been forwards compatible. It used to be very noticable in timing routines, running Dune 2 on a fast 386, or, god forbid, a 486.
(For the Robot Odyssey example, play DroidQuest instead. It's a Java port of RO.)
I expect to see a lot of similar problems with OSS software. Not because of timing routines, but because of unclean code. For those who are accustomed to writing 32-bit code, making and keeping that code 64-bit clean is going to be a concious effort.