In the INTERPRETATION of the normal GPL someone decided that "derived work" would include linked-binaries. Even if say you just included a verbatim copy of a certain GPL library function.
The intent of GPL is that you get the source, and are free to modify the source. If you do so, and distribute the result, you also get to distribute your modifications to the original source.
In the case of the font, do you use the source of the font to modify it a little bit, and distribute the results?
If you write a spreadsheet using gnumeric, does the spreadsheet become GPL? Ah, it's a separate file? but what if you print-to-ps an extract of the spreadsheet and intend to distribute that? (but not your formulas). Now your spreadsheet is sort of "linked" with the GPL stuff from gnumeric into one file. Is the separate "source" (your spreadsheet) suddenly "derived" from the gnumeric source? Come on!
What prompted the GPL was that the BSD licence allowed SUN and DEC to take BSD, fix bugs, and then sell the result. This meant that they both had to fix the same bugs separately from the "open source" people. This is a waste of time. DEC and SUN clearly had a "derived" product.
If I make a program that does something, but requires say a CRC32. I can find a GPL CRC32 source on the internet, and I'd argue that as long as I don't modify a byte in the crc32.c file, my whole directory with sources isn't suddenly a derived work of the crc32.c that I got from the internet. Similarly, if I compile it and happen to statically link the binary, all of my sources are not suddenly derived from the crc32.c I found on the internet. Now the binary still DOES contain GPL stuff. So I get to provide the LICENCE file with my software, and I need to make the crc32.c source available to anyone who asks.
Suppose I write proprietary software. Suppose I distribute it on a CD, and I decide to put a small GPL utility on the CD as well. These two are "linked" on the cd.img.iso image, right? Suddenly my million lines of code are "derived" from that simple GPL utility source code? Get reasonable!
Of course people buying the CD get the right to the source of the utility.
Of course people getting a "linked" version of your document with GPL fonts get the right to the source of the GPL font. But they don't automatically get the right to publish (modified or not) versions of your document.
I see lots of reactions that suggest to buy an IBM or other "High quality" computer. ("For under $30k").
But this is not the issue. The idea behind the question is that RAM now costs some EUR 120 per Gb in my local store, so for EUR 1000,- I can buy 8G of RAM. Spend another EUR 1000,- and I can have an 8G machine, right? (For americans, read dollars where I say EUR.)
Well, it turns out that this is difficult. Modern CPUs have the memory controller integrated in the CPU and it is not easy to make a motherboard that can address more than the standard amount per CPU.
This means that the motherboard is not going to be available for say "under EUR 1000".
It's interesting you think that, because if what you describe were true then Microsoft would be continuing a licensing practice in violation of anti-trust laws for which they were found "guilty".
Not being a vendor myself, I can't be sure. Sorry.
However, even in the face of lost anti-trust-suits MicroSoft would like to continue this practise, so if they are no longer allowed to put the "all PCs" clause in the contract, they will try something like: We'll give you 30% extra discount if you allow us to charge you for "PCs sold" instead of "PCs sold with Windows".
They can always argue that this is because the accounting is easier and more reliable, so that they are doing this to improve direct income.
The thing is, microsoft offers the vendor $99 (i.e. a lower price) per laptop in licencing fees, provided that the vendor ships ALL laptops with the Microsoft OS. In any case, they don't pay per laptop shipped with the Microsoft OS, they pay per laptop shipped. Even yours after you refuse to pay for the OS you won't use.
So they could just refund you your $99, and leave it at that. Just tell you: Please don't use it.
However that story would quickly get out, and be misused.
The thing is, if they sell 1000 laptops with the Microsoft OS, microsoft will emphasize that they are offering the licence at say $99 per copy. The company should realize that they are getting a reasonable deal at $100 per copy for the 990 users that DO want Windows installed.
I'm not totally familiar with the US rules, but by the time you can financially survive the "$1 salary" situation, you're probably already paying much more than average in taxes through other routes.
FYI, here in the Netherlands (aka Holland), you ARE allowed to pay yourself just little salary should the circumstances dictate this.
It is of course NOT allowed to say have the company show more than half a salary in profit, and then not to pay anything to the boss. But if you say show a profit of 1 tenth (or less) of a normal salary, it can be argued that dishing out salary to the guy in charge would be commercial suicide.
In Holland (aka the Netherlands), the "boss" has to earn a minimum amount. This is quite a lot higher than minimum wage. In fact he also has to earn more than the best-paid employee.
This is indeed meant to prevent the situation described above where "nil paid income" means "little income taxes", where the boss covers his immediate expenses with sold stock (if you're the boss/significant shareholder that counts as income here as well!) or previous income.
As you can see, this is a TAX rule, so it's the tax department who gets to complain, not the employee.
The registers are not the problem. If you take an SRAM chip designed for 2.5V and lower the voltage a bit, and then bring it up to 2.5V to test-read the contents of the cells, you'll find that you can go VERY VERY far. In practise, removing the power completely and then bringing it back will show significantly more than 50% of the cells still having the right contents.
There are two (related) problems: When a combinatorial part of the processor does its thing, it slows down if it gets a lower voltage. So the results are available later. If that is after the next clock, the wrong results will be clocked into the next register.
Similarly, if you have a long line going from one end of the chip to another, this will slow down, leading to the results not getting there in time.
These problems are easily fixed: use the right voltage.
Once you drop the speed enough to allow lowering the voltage, you can also do that.
When you lower the clock speed, the power the chip produces goes down linearly. If however, you lower the clockspeed, AND lower the voltage on which the core of the CPU runs (which is what speedstep and the likes do!) then power consumption goes down even faster.
Last time I tried that, the motherboard was not "designed for overclockers" and wouldn't allow me to do this.:-(
if you're in a public area it's fair game to be photographed
It seriously depends on what culture you're from. I was in marocco over christmas, and over there some percentage of people have religious beliefs that their photograph should NOT be taken.
So, us westerners take this for granted, while others don't.
Here in the Netherlands, if you drive a car, your licence plate is displayed. However, if the authorities start taking pictures of licence plates on public roads, suddenly we get protests because this would infringe on privacy.
Taking pictures and analysing them by computer, recognizing people is suddenly something that lots of people seem to accept. Still it carries the same or worse privacy issues than the licence plates. Boundaries differ. In time and in place.
Backups are easy if your content is less than what fits on cheap removable media.
Note that backing up on removable media usually requires human intervention: insert the ipod, enter a blank CD. This means that by the time your harddrive finally dies, you'll find you didn't make a backup for way too long.
I sometimes type "mplayer *.vob", and get all the exta's/hidden content without having to mess around....
Re:It's scarey they think that is a solution
on
Tin Foil Passports?
·
· Score: 1
I am (an european and..) by law not allowed to hand over my passport unless there is a legal duty to do so.
Most hotels honor my: "Sure you can make a copy, but I have to keep my passport on me". That's enough of a middle ground that it is usually accepted as a solution.
Of course, they have to do business with me. So if I fail to comply with their house-rules, they can refuse my business.
Of course, if an artist sticks to the same techniques during his lifetime, he'll be identified. If a "known limitation" of the method is that it can only compare paintings of the same type, then you still risk the possibility that say Rembrandt decided to paint a picture with his left hand instead of his right because he hit his right thumb with a hammer while fixing his house. This will falsely label a picture as a fake.
Similarly, in the time of Rembrandt, it was common to have helpers paint the less important parts of the picture. What if Rembrandt switches apprentices?
Policy also probably says that you can't use your work computer for anything but work,
OK. I am the guy making the policies here, and my opinion, people are more effective if they are having fun as well. If you enjoy a moment of "time off" reading slashdot, that's fine with me. If you just have to do some quick banking things from work, fine.
and unless you happen to be the finance person checking the company account, you shouldnt be doing your banking at work, sure everyone does it, but in a contract/liabilty sense - you werent supposed to.
It's a small company. I'm the finance person doing the company banking things. Now what?
The thing is, these "controls" are there for a reason. Turning them off disallows these scripts from taking over your computer. But apparently there are some websites or somesuch that simply use the features provided by this type of script. As their computer won't allow some functionality they ejoy, people are always tempted to turn these things on.
When I was a student, I did my internship at Philips. They were working on this back then. In 1990. "Almost ready". "expect results on the market in a year or two".... Yeah right.
Hardare raid has reasonable performance, works as it should in simple cases, but when the shit really hits the fan, you're totally screwed.
As a hardware RAID array doesn't have a console, and no option to "ask questions" to the user, it has to automatically DO things as drives are replaced.
With Linux software raid, you end up with the option of TELLING linux what the config is, and possibly recovering some part of your data, should things go seriously wrong. In a hardware RAID, you're totally dependent on the firmware in the card to "do the right thing".
Suppose you configure 7 drives as a RAID, and the 8th as a hot spare. The firmware will mark the drives with a config version and drive number:
1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7 and hot spare 1.0
Next drive 3 breaks.
1.1, 1.2, XXX, 1.4, 1.5, 1.6, 1.7 and hot spare 1.0
and the stuff is reconfigured:
2.1, 2.2, XXX, 2.4, 2.5, 2.6, 2.7 and 2.3
If now the person handling the drives makes a mistake and removes the 4th drive: 2.4 instead of the the third:
3x.1, 3x.2, XXX, 0.0, 3x.5, 3x.6, 3x.7 and 3x.3
so now whe have a broken raidconfig: "3x". No amount of replacing 2.4 will make it fit in, as the array has determined that the raid has been broken.
If you get this far, "datarecvery" is the only thing that can help. Don't write a byte on any drive and try to get things right again. Drives that can be jumpered to "readonly" would be a help. We comercially would just make an image of all the drives and work from there....
Due to the process, higher DPI numbers are used in printers and printing. Your average analog photos are nowadays digitally transferred to photopaper, at 300 DPI. So the 5k x 3.2k pixels will allow you to blow up your image to a 16x10 inch poster without starting to see the grain. Blow it up to 20x32, and you still have excellent quality.
Note that my 5Mpixel camera already takes nicer pictures than most 35mm analog cameras. The artefacts are however different, so that the analog freaks claim that their stuff makes better pictures.
The really annoying thing is that it's often wrong. They once tested with a different browser and now have a javascript check my browser and redirect me away even if my browser displays correctly.
The good way to handle this is to just make a page that displays good to reasonable on all browsers.
The reasonable way to handle it is to issue a warning: "We haven't tested our website with your browser, we don't have the resources to fix it if it doesn't display correctly, you're on your own if that happesn to be the case".
I hate it when I'm "denied access" just because the string "galeon" doesn't match "IE".
See http://www.openbrick.com when it works again. If I remember correctly, they draw about 1 to 2A, or about 5-10W. This was measured without a harddisk.
Even if you have a 3000x2000 desktop, your average DVD decodes to 800x480. THAT is the resolution of the video-data that needs transporting to the video card. The video card should have accelleration to bit-blit that out to full-screen video.
The standard trick that is used is to tell the 3D chip that we're looking at a wall that has a texture and change the texture 50 times a second.
Obviously you want dual caches. Even if a single shared cache would be a benefit with a slightly higher cache hit rate, you have two 3GHz plus cores bashing at it at > 3 billion times a second. In that case you don't want to deal with the cores tying up each other for ALL cache-accesses that they do.
It is certainly worth it, to push the point where memory hierachies meet as far as possible from the core. That's why you get big caches in the XEONs meant for multiprocessing.
Now, if a dual (triple?) ported cache were "free", then of course you'd go for the dual ported version: double the cache size, and your hit ratio goes up a little. But they are not free. You pay in speed, area, transistor count etc. In the end, I'm pretty sure Intel and AMD will have evaluated the tradeoffs and taken the right decision.
With a bit of luck, they have a "fastpath" between the caches for stuff that needs to bounce around between the two caches/cores.
Slashdot Mirror
In the INTERPRETATION of the normal GPL someone decided that "derived work" would include linked-binaries. Even if say you just included a verbatim copy of a certain GPL library function.
The intent of GPL is that you get the source, and are free to modify the source. If you do so, and distribute the result, you also get to distribute your modifications to the original source.
In the case of the font, do you use the source of the font to modify it a little bit, and distribute the results?
If you write a spreadsheet using gnumeric, does the spreadsheet become GPL? Ah, it's a separate file? but what if you print-to-ps an extract of the spreadsheet and intend to distribute that? (but not your formulas). Now your spreadsheet is sort of "linked" with the GPL stuff from gnumeric into one file. Is the separate "source" (your spreadsheet) suddenly "derived" from the gnumeric source? Come on!
What prompted the GPL was that the BSD licence allowed SUN and DEC to take BSD, fix bugs, and then sell the result. This meant that they both had to fix the same bugs separately from the "open source" people. This is a waste of time. DEC and SUN clearly had a "derived" product.
If I make a program that does something, but requires say a CRC32. I can find a GPL CRC32 source on the internet, and I'd argue that as long as I don't modify a byte in the crc32.c file, my whole directory with sources isn't suddenly a derived work of the crc32.c that I got from the internet. Similarly, if I compile it and happen to statically link the binary, all of my sources are not suddenly derived from the crc32.c I found on the internet. Now the binary still DOES contain GPL stuff. So I get to provide the LICENCE file with my software, and I need to make the crc32.c source available to anyone who asks.
Suppose I write proprietary software. Suppose I distribute it on a CD, and I decide to put a small GPL utility on the CD as well. These two are "linked" on the cd.img.iso image, right? Suddenly my million lines of code are "derived" from that simple GPL utility source code? Get reasonable!
Of course people buying the CD get the right to the source of the utility.
Of course people getting a "linked" version of your document with GPL fonts get the right to the source of the GPL font. But they don't automatically get the right to publish (modified or not) versions of your document.
I see lots of reactions that suggest to buy an IBM or other "High quality" computer. ("For under $30k").
But this is not the issue. The idea behind the question is that RAM now costs some EUR 120 per Gb in my local store, so for EUR 1000,- I can buy 8G of RAM. Spend another EUR 1000,- and I can have an 8G machine, right? (For americans, read dollars where I say EUR.)
Well, it turns out that this is difficult. Modern CPUs have the memory controller integrated in the CPU and it is not easy to make a motherboard that can address more than the standard amount per CPU.
This means that the motherboard is not going to be available for say "under EUR 1000".
It's interesting you think that, because if what you describe were true then Microsoft would be continuing a licensing practice in violation of anti-trust laws for which they were found "guilty".
Not being a vendor myself, I can't be sure. Sorry.
However, even in the face of lost anti-trust-suits MicroSoft would like to continue this practise, so if they are no longer allowed to put the "all PCs" clause in the contract, they will try something like: We'll give you 30% extra discount if you allow us to charge you for "PCs sold" instead of "PCs sold with Windows".
They can always argue that this is because the accounting is easier and more reliable, so that they are doing this to improve direct income.
You're assuming the licence costs say $100.
The thing is, microsoft offers the vendor $99 (i.e. a lower price) per laptop in licencing fees, provided that the vendor ships ALL laptops with the Microsoft OS. In any case, they don't pay per laptop shipped with the Microsoft OS, they pay per laptop shipped. Even yours after you refuse to pay for the OS you won't use.
So they could just refund you your $99, and leave it at that. Just tell you: Please don't use it.
However that story would quickly get out, and be misused.
The thing is, if they sell 1000 laptops with the Microsoft OS, microsoft will emphasize that they are offering the licence at say $99 per copy. The company should realize that they are getting a reasonable deal at $100 per copy for the 990 users that DO want Windows installed.
I'm not totally familiar with the US rules, but by the time you can financially survive the "$1 salary" situation, you're probably already paying much more than average in taxes through other routes.
FYI, here in the Netherlands (aka Holland), you ARE allowed to pay yourself just little salary should the circumstances dictate this.
It is of course NOT allowed to say have the company show more than half a salary in profit, and then not to pay anything to the boss. But if you say show a profit of 1 tenth (or less) of a normal salary, it can be argued that dishing out salary to the guy in charge would be commercial suicide.
In Holland (aka the Netherlands), the "boss" has to earn a minimum amount. This is quite a lot higher than minimum wage. In fact he also has to earn more than the best-paid employee.
This is indeed meant to prevent the situation described above where "nil paid income" means "little income taxes", where the boss covers his immediate expenses with sold stock (if you're the boss/significant shareholder that counts as income here as well!) or previous income.
As you can see, this is a TAX rule, so it's the tax department who gets to complain, not the employee.
This is complete bullshit. (pardon my french).
The registers are not the problem. If you take an SRAM chip designed for 2.5V and lower the voltage a bit, and then bring it up to 2.5V to test-read the contents of the cells, you'll find that you can go VERY VERY far. In practise, removing the power completely and then bringing it back will show significantly more than 50% of the cells still having the right contents.
There are two (related) problems: When a combinatorial part of the processor does its thing, it slows down if it gets a lower voltage. So the results are available later. If that is after the next clock, the wrong results will be clocked into the next register.
Similarly, if you have a long line going from one end of the chip to another, this will slow down, leading to the results not getting there in time.
These problems are easily fixed: use the right voltage.
Once you drop the speed enough to allow lowering the voltage, you can also do that.
When you lower the clock speed, the power the chip produces goes down linearly. If however, you lower the clockspeed, AND lower the voltage on which the core of the CPU runs (which is what speedstep and the likes do!) then power consumption goes down even faster.
:-(
Last time I tried that, the motherboard was not "designed for overclockers" and wouldn't allow me to do this.
if you're in a public area it's fair game to be photographed
It seriously depends on what culture you're from. I was in marocco over christmas, and over there some percentage of people have religious beliefs that their photograph should NOT be taken.
So, us westerners take this for granted, while others don't.
Here in the Netherlands, if you drive a car, your licence plate is displayed. However, if the authorities start taking pictures of licence plates on public roads, suddenly we get protests because this would infringe on privacy.
Taking pictures and analysing them by computer, recognizing people is suddenly something that lots of people seem to accept. Still it carries the same or worse privacy issues than the licence plates. Boundaries differ. In time and in place.
Where does that leave me? I have 35Gb of photos.
Backups are easy if your content is less than what fits on cheap removable media.
Note that backing up on removable media usually requires human intervention: insert the ipod, enter a blank CD. This means that by the time your harddrive finally dies, you'll find you didn't make a backup for way too long.
Two friends of mine "competed" against each other. After a month the score was 15-all, and they called a truce.
They found 15 NEW Unix exploits each! Of course there were rules. For example: AIX was "off limits" (too easy).
Now, of course, this was a long time ago, and buffer overflows were not popular.
Apparently DJB just achieved 11% (44 out of 400) of his goals. I think he should fail his course.....
I sometimes type "mplayer *.vob", and get all the exta's/hidden content without having to mess around....
I am (an european and..) by law not allowed to hand over my passport unless there is a legal duty to do so.
Most hotels honor my: "Sure you can make a copy, but I have to keep my passport on me". That's enough of a middle ground that it is usually accepted as a solution.
Of course, they have to do business with me. So if I fail to comply with their house-rules, they can refuse my business.
Of course, if an artist sticks to the same techniques during his lifetime, he'll be identified. If a "known limitation" of the method is that it can only compare paintings of the same type, then you still risk the possibility that say Rembrandt decided to paint a picture with his left hand instead of his right because he hit his right thumb with a hammer while fixing his house. This will falsely label a picture as a fake.
Similarly, in the time of Rembrandt, it was common to have helpers paint the less important parts of the picture. What if Rembrandt switches apprentices?
Policy also probably says that you can't use your work computer for anything but work,
OK. I am the guy making the policies here, and my opinion, people are more effective if they are having fun as well. If you enjoy a moment of "time off" reading slashdot, that's fine with me. If you just have to do some quick banking things from work, fine.
and unless you happen to be the finance person checking the company account, you shouldnt be doing your banking at work, sure everyone does it, but in a contract/liabilty sense - you werent supposed to.
It's a small company. I'm the finance person doing the company banking things. Now what?
The thing is, these "controls" are there for a reason. Turning them off disallows these scripts from taking over your computer. But apparently there are some websites or somesuch that simply use the features provided by this type of script. As their computer won't allow some functionality they ejoy, people are always tempted to turn these things on.
When I was a student, I did my internship at Philips. They were working on this back then. In 1990. "Almost ready". "expect results on the market in a year or two". ... Yeah right.
Hardare raid has reasonable performance, works as it should in simple cases, but when the shit really hits the fan, you're totally screwed.
As a hardware RAID array doesn't have a console, and no option to "ask questions" to the user, it has to automatically DO things as drives are replaced.
With Linux software raid, you end up with the option of TELLING linux what the config is, and possibly recovering some part of your data, should things go seriously wrong. In a hardware RAID, you're totally dependent on the firmware in the card to "do the right thing".
Suppose you configure 7 drives as a RAID, and the 8th as a hot spare. The firmware will mark the drives with a config version and drive number:
1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7 and hot spare 1.0
Next drive 3 breaks.
1.1, 1.2, XXX, 1.4, 1.5, 1.6, 1.7 and hot spare
1.0
and the stuff is reconfigured:
2.1, 2.2, XXX, 2.4, 2.5, 2.6, 2.7 and 2.3
If now the person handling the drives makes a mistake and removes the 4th drive: 2.4 instead of the the third:
3x.1, 3x.2, XXX, 0.0, 3x.5, 3x.6, 3x.7 and 3x.3
so now whe have a broken raidconfig: "3x". No amount of replacing 2.4 will make it fit in, as the array has determined that the raid has been broken.
If you get this far, "datarecvery" is the only thing that can help. Don't write a byte on any drive and try to get things right again. Drives that can be jumpered to "readonly" would be a help. We comercially would just make an image of all the drives and work from there....
Due to the process, higher DPI numbers are used in printers and printing. Your average analog photos are nowadays digitally transferred to photopaper, at 300 DPI. So the 5k x 3.2k pixels will allow you to blow up your image to a 16x10 inch poster without starting to see the grain. Blow it up to 20x32, and you still have excellent quality.
Note that my 5Mpixel camera already takes nicer pictures than most 35mm analog cameras. The artefacts are however different, so that the analog freaks claim that their stuff makes better pictures.
The really annoying thing is that it's often wrong. They once tested with a different browser and now have a javascript check my browser and redirect me away even if my browser displays correctly.
The good way to handle this is to just make a page that displays good to reasonable on all browsers.
The reasonable way to handle it is to issue a warning: "We haven't tested our website with your browser, we don't have the resources to fix it if it doesn't display correctly, you're on your own if that happesn to be the case".
I hate it when I'm "denied access" just because the string "galeon" doesn't match "IE".
the question now is: "what, if anything, should be done with it?"
Well, when the explosives squad finds a WW2 bomb out here, they tag on some extra explosives and just make it go boom.
That's one I'd like to watch..... From a different planet.
See http://www.openbrick.com when it works again. If I remember correctly, they draw about 1 to 2A, or about 5-10W. This was measured without a harddisk.
Even if you have a 3000x2000 desktop, your average DVD decodes to 800x480. THAT is the resolution of the video-data that needs transporting to the video card. The video card should have accelleration to bit-blit that out to full-screen video.
The standard trick that is used is to tell the 3D chip that we're looking at a wall that has a texture and change the texture 50 times a second.
Obviously you want dual caches. Even if a single shared cache would be a benefit with a slightly higher cache hit rate, you have two 3GHz plus cores bashing at it at > 3 billion times a second. In that case you don't want to deal with the cores tying up each other for ALL cache-accesses that they do.
:-(
It is certainly worth it, to push the point where memory hierachies meet as far as possible from the core. That's why you get big caches in the XEONs meant for multiprocessing.
Now, if a dual (triple?) ported cache were "free", then of course you'd go for the dual ported version: double the cache size, and your hit ratio goes up a little. But they are not free. You pay in speed, area, transistor count etc. In the end, I'm pretty sure Intel and AMD will have evaluated the tradeoffs and taken the right decision.
With a bit of luck, they have a "fastpath" between the caches for stuff that needs to bounce around between the two caches/cores.
Oh. I haven't read the article. Sorry.