That was George Pal's War of the Worlds, made in 1953. George Pal also made the other movie you mentioned, The Time Machine in 1960.
George Pal also made When Worlds Collide and several other big budget (for the time) science fiction films. He really was the Spielberg/Lucas of that time.
Should I bother pointing out the Slashdot did in fact put an article about the doubts about that papers on here on Friday? (there were two articles, but the later one was specifically about the fact that many people doubt the authenticity of the documents)
No, you just want to continue to live with your delusions. No fact is going to change your opinion...
The column was funny, but I agree that it may lead more people to misunderstand the problem. The joke about the hacker is funny but misleading. It would help if Barry had added to his description of how it works "and then the machine adds one to Boba Fett, because that is the guy who programmed the machine's favorite Star Wars character".
Worries about "hackers" are used to hide the true problems with voting machines.
I just wanted to pick one nit with your post. The GPL does not use copyright against copyright, it uses copyright to attack proprietary secret code, which has nothing to do with copyright (secrets don't have to be copyrighted). You may be confusing GPL with "fair use". "Fair use" which has lots of support here, is probably the only possible hole in the GPL, somebody can maybe claim that using a small enough piece of GPL code should be allowed by "fair use".
As for the original question, it may be possible to say that your patented invention is free for use for GPL code, but you must buy a license for any other use. And then price the license so high that it cannot be used except for GPL. As another poster said it may not be possible to explicitly limit some uses of the idea.
Um, 1984 was clearly aimed at Communist Dictatorships. I don't like Bush either, but that does not mean everything in the world is a description of their problems.
Brave New World was the contemporary opposite attack on consumerism.
Tables look pretty good in an old Konqueror. I can see where he wants the vertical text, but to be honest I think the horizontal version I got is more readable.
As I pointed out in another letter, the "test image" certainly does not line up. The starts of words do, but the ends do not. Words get larger while the spaces between them get smaller. I can think of no plausable way a fax could distort the message in such a way.
Reasons for the lineup? It is possible that the person making the lineup microspaced it on purpose to line it up. If so, shame on you, you have completely botched up any argument by making your own forgery.
Another possibility is that Microsoft Word is moving words to match the on-screen pixel positions, and that the original document is Word output, like claimed. It is possible that both were composed with the same font but printed with different printers, one of them PostScript with it's own, slightly different, internal font. It seems likely it is the LGF document, as it seems the spacing is much more irregular there.
I have seen and even typed on a proportionally-spaced typewriter of that era. Now I was 10 at the time and only saw it once or twice, and also saw plenty of fixed-pitch IBM Selectrics so I may be confused. But I don't believe this is output from such a typewriter now. Most of the other arguments are legitimate: the "th", though it could be a letter on the typeball, looks too much like a scaled-down one, back then it would be hand-drawn and probably squashed horizontally. Also the typewriters had only a limited number of widths (I think numbers were 3 units) and this meant matching horizontal positions between lines were much more common than is visible here. The fonts also looked much more like typewriters, often only half the capital letters and a few lower-case ones were modified. I also believe the lower-case L and the upper-case O had to be the same width as numbers so they could be used as 1 and 0, all numbers must have the same width or it is impossible to line up columns. This is quite evidently not true for the lower-case L in the documents.
If you look at the test, it is pretty obvious that the start of words line up much better than the ends of words, except at the ends of lines where it is the opposite, or the second-to-last word which are centered. I suspect this is because this is a purposeful attempt to match by using horizontal adjustment, or possibly the supplied Word document does not print the image shown. If it actually matches and does not have any Word microspacing then it is an absolutely amazing coincidence. There is no possible results of faxing and copying that would move, for instance, the end of the word "obviously" while not moving the start of it and not moving the start of the word "pressured" a similar amount.
It does look like the test image has a.6 scaled version of the PDF fax file without adjustment, so I belive the adjustment was done in the Word document or to it's output, and not to the PDF. I believe this was done by using Word's microspacing, though maybe some crazy person just adjusted the image using Photoshop or something...
I have no idea how to see the microspacing or adjust it, and I don't have a copy of Word anyway (it opens in AbiWord though). If the word document contains no microspacing and actually produces a matching image like this, it is a coincidence of astronomical proportions, I think. It is pretty obvious that they have not reproduced the original, but that does not mean the original is real.
Seperate the channels and look at the "Word" output (red channel) and it appears the words have been moved horizontally to line them up with the original fax. Notice that the first letter in each word lines up much better than the last one, and that the spaces seem to be different sizes between the words (compare "Harris gave" to "gave me" for instance). In addition the blue channel is not the same resolution as the pdf file, it appears to have been scaled down by a non-integer, which blurs and thickens the original letters. Also they put red*blue into the green channel to make it look more consistent.
However I have to say that zooming in on the original PDF documents does reveal some strange artifacts. They seem to have been a 3 or 4 level image, with very even gray fringes around the letters. The "th" is also questionable, though there certainly were symbol balls for the Selectric I'm not really certain if it included a small th that looks so exactly scaled. I would also question if anybody would really title the thing "CYA". Because of these doubts it is rather foolish of somebody to make bogus tests when there may be real proof coming, it will only discredit you.
My quick opinion is that if this test is real then these certainly are forgeries, produced just as he said.
Though proportional-spaced typewriters existed then and were quite common (despite claims to the contrary by some people here), they were still mechanical devices. There were only 4 (perhaps 5) letter widths possible, and the numbers (the "en-space") were 3 units (this is for the IBM Selectric I am familiar with). This produces obvious alignement vertically between far more letters than the Word output, as there is a 1/3 chance of a letter aligning with one below it. TrueType fonts have 1000 or more possible widths (I may be thinking of PostScript Type 1 fonts).
It is extremely unlikely that a TrueType font matches the widths used by the Selectric for every letter, too.
Since the documents were covered with dirt and distortions, that would indicate an attempt to make a forgery. No excuse about somebody retyping is going to fly.
PS: I'm going to vote for Kerry anyway. Still somebody is being made a fool here. CBS is really going to lose if these are fake, even retracting their claim is not going to help. The Dems can get out of this easily by saying somebody tried to discredit them with fake documents.
Re:Why no comparison with D3D?
on
OpenGL 2.0 Released
·
· Score: 2, Informative
What? I'm sure there are legitimate arguments for D3D, but saying the C-API of OpenGL makes it "work worse with other languages" is about the most idiotic thing I have ever seen. It is well known that one of the major reasons for limiting an API to C is to make it callable from multiple languages, ie the exact opposite!
I would be much more convinced with an argument that D3D "rejects old languages and does not compromise the interface in order to support them" which is a legitimate argument.
IBM Selectric typewriters had a single back-quote character, and a single forward quote that was also used as apostrophe. I seem to remember that they also had double back and double forward quote characters, but those may have been missing on the proportional-spaced model that I also remember somewhat. The certainly did not have the neutral double quote from ASCII. Early ASCII teletypes also had a neutral single quote, it was changed on newer systems that added the backquote as well as the lower-case characters.
Selectric typewriters also had 1/2 and some other fractions, a copyright symbol, and some others. They did not have curly braces, less/greater, and many other ASCII symbols. You could also change the ball to a "symbol" ball that had greek (this may not have been possible on the proportional models?)
Such machines existed in the early 70's. My mom used one. I'm going to stop repeating this, because the people claiming the forgeries have made asses of themselves by now.
Proportional-spaced IBM Selectric typewriters (and perhaps other brands) existed in the early 70's, and probably long before. My mom used one, and I played with it trying to make pictures on the paper (the spacing offered the chance to make much fancier graphics, but the machine she was using lacked any way to advance by 1 unit, which limited the ability to place things where wanted.) The machine looked exactly like the fixed-spacing machine she had at home, but I don't believe one could be altered to the other, the character widths were hard-coded.
Before you make an ass out of yourself, you might want to check out IBM selectric typewriters from that time. My mom used one of these. If I remember right, there were 4 or perhaps more different character widths, numbers and "en space" were 3. One frustration to a little proto-nerd kid trying to draw pictures on this was that there was no "move by 1 unit" key.
Actually this new system is exactly what you want. The windows have an alpha channel (they are 4-channel off-screen images). These images are handed to the compositing manager, not back to your program, which puts them on the screen.
The only difference from what you want is that the compositing manager can be replaced and can interpret the alpha in other ways besides just stacking the windows. This means that the drop shadows can be added at the compositing step, rather than just making the window bigger with a partially-transparent gray edge. Also means things like fading out the background windows can be done (OS/X obviously redraws the windows to fade the inactive ones).
Running older X11 applications, including ones from the Sun workstation, will work.
The opposite is not true, programs that use these extensions on Linux will not work on remote X servers without them. Now, in theory, programs are supposed to test for the extensions and work around them if they are missing, but I know in reality it is much easier to just assume the extension is there, or to quit with an error if it is not.
In fact the desire to have programs work without the extension has led to much of the stagnation of X. Any complicated extension has been rejected because emulating it is much more work than just not using it at all.
It would help a lot if the authors would produce a library that uses the extension and also provides the emulation if the extension is missing. This was done in Xft and is probably the reason why it was adopted in what seemed like weeks by every piece of Linux software. But it is a lot of work so I don't expect this to be provided for these extensions.
Several of the responders seem confused here. What is wanted is not to literally put every detail of this into the kernel, no more than the details of ext2 are in the base kernel code. Nor does this mean that the entire database interface must be in the kernel.
What is needed is the ability to take any selected object, anywhere, identify it with a "filename" that is a null-terminated string of bytes, and send it to open() and get a link to that object. Then when that SMB selection is made, it can be sent to OpenOffice and it will, guaranteed, open it! The filename can be complete gibberish, OpenOffice does not have to understand it, it just has to rely that it can open it and get the same thing the user expected.
I don't understand why anybody thinks we want to use any interface other than open() to get at objects. It seems painfully obvious to me, yet both Gnome and KDE seem to think we need to learn complex, incompatable interfaces.
It also does not help that there are morons out there pushing "case independence" and "wide characters" and i18n using anything other than utf-8. All of these are a serious impediment to getting this done correctly.
It would help a lot if tar would do it if you just provided -z instead of having to remember to provide -j. Come to think of it, it would be nice if tar just detected compression and you did not have to give it -z either! Can this be done?
The problem was with code that checked if entries in the matrix were zero to avoid degenerate cases that did not work with such rotations. One instance was code to halve a 4x4 perspective transform into two equal transforms. The very small numbers produced math overflows. I did worry about real rotations near 90 producing such small numbers, but it was not possible to produce such small numbers from values of 90 +/- D where D is the minimum delta of a double at a value of 90. Therefore the only way to get it was due to the error between the representable value of M_PI_2 and the perfect value.
Certainly like you say the code needed to be fixed. My post was to point out that the correct fix is to use matricies at a much higher level, abandoning angles as soon as possible. Quaternions are also useful.
Right-angle rotations are betters specified as matricies, where all the coefficients are 1 and 0. I have certainly been burned by math errors where exactly 90 turned into approximatly pi/2 and that turned into 1e-9 in the cos() and this error was later magnified to completely wrong output.
Also use M_PI_2 if you can in your code (of coure Microsoft did not put that in their math.h header file, copy it from the linux header file). I would expect any 3D language to provide a symbol that produces that value with maximum accuracy.
All computer companies say that. It apparently gets them a discount on the Windows bulk pricing.
Quite likely if Linux takes off and they want to buy it from RedHat and RedHat gave them a discount for doing so, they would print "HP recommends RedHat Linux" right next to the Microsoft ad, and look even more silly.
Slashdot Mirror
That was George Pal's War of the Worlds, made in 1953. George Pal also made the other movie you mentioned, The Time Machine in 1960.
George Pal also made When Worlds Collide and several other big budget (for the time) science fiction films. He really was the Spielberg/Lucas of that time.
Any unexpected cleanup of the plane interior after the flight may cost you extra.
Should I bother pointing out the Slashdot did in fact put an article about the doubts about that papers on here on Friday? (there were two articles, but the later one was specifically about the fact that many people doubt the authenticity of the documents)
No, you just want to continue to live with your delusions. No fact is going to change your opinion...
The column was funny, but I agree that it may lead more people to misunderstand the problem. The joke about the hacker is funny but misleading. It would help if Barry had added to his description of how it works "and then the machine adds one to Boba Fett, because that is the guy who programmed the machine's favorite Star Wars character".
Worries about "hackers" are used to hide the true problems with voting machines.
I just wanted to pick one nit with your post. The GPL does not use copyright against copyright, it uses copyright to attack proprietary secret code, which has nothing to do with copyright (secrets don't have to be copyrighted). You may be confusing GPL with "fair use". "Fair use" which has lots of support here, is probably the only possible hole in the GPL, somebody can maybe claim that using a small enough piece of GPL code should be allowed by "fair use".
As for the original question, it may be possible to say that your patented invention is free for use for GPL code, but you must buy a license for any other use. And then price the license so high that it cannot be used except for GPL. As another poster said it may not be possible to explicitly limit some uses of the idea.
Um, 1984 was clearly aimed at Communist Dictatorships. I don't like Bush either, but that does not mean everything in the world is a description of their problems.
Brave New World was the contemporary opposite attack on consumerism.
Tables look pretty good in an old Konqueror. I can see where he wants the vertical text, but to be honest I think the horizontal version I got is more readable.
As I pointed out in another letter, the "test image" certainly does not line up. The starts of words do, but the ends do not. Words get larger while the spaces between them get smaller. I can think of no plausable way a fax could distort the message in such a way.
Reasons for the lineup? It is possible that the person making the lineup microspaced it on purpose to line it up. If so, shame on you, you have completely botched up any argument by making your own forgery.
Another possibility is that Microsoft Word is moving words to match the on-screen pixel positions, and that the original document is Word output, like claimed. It is possible that both were composed with the same font but printed with different printers, one of them PostScript with it's own, slightly different, internal font. It seems likely it is the LGF document, as it seems the spacing is much more irregular there.
I have seen and even typed on a proportionally-spaced typewriter of that era. Now I was 10 at the time and only saw it once or twice, and also saw plenty of fixed-pitch IBM Selectrics so I may be confused. But I don't believe this is output from such a typewriter now. Most of the other arguments are legitimate: the "th", though it could be a letter on the typeball, looks too much like a scaled-down one, back then it would be hand-drawn and probably squashed horizontally. Also the typewriters had only a limited number of widths (I think numbers were 3 units) and this meant matching horizontal positions between lines were much more common than is visible here. The fonts also looked much more like typewriters, often only half the capital letters and a few lower-case ones were modified. I also believe the lower-case L and the upper-case O had to be the same width as numbers so they could be used as 1 and 0, all numbers must have the same width or it is impossible to line up columns. This is quite evidently not true for the lower-case L in the documents.
If you look at the test, it is pretty obvious that the start of words line up much better than the ends of words, except at the ends of lines where it is the opposite, or the second-to-last word which are centered. I suspect this is because this is a purposeful attempt to match by using horizontal adjustment, or possibly the supplied Word document does not print the image shown. If it actually matches and does not have any Word microspacing then it is an absolutely amazing coincidence. There is no possible results of faxing and copying that would move, for instance, the end of the word "obviously" while not moving the start of it and not moving the start of the word "pressured" a similar amount.
.6 scaled version of the PDF fax file without adjustment, so I belive the adjustment was done in the Word document or to it's output, and not to the PDF. I believe this was done by using Word's microspacing, though maybe some crazy person just adjusted the image using Photoshop or something...
It does look like the test image has a
I have no idea how to see the microspacing or adjust it, and I don't have a copy of Word anyway (it opens in AbiWord though). If the word document contains no microspacing and actually produces a matching image like this, it is a coincidence of astronomical proportions, I think. It is pretty obvious that they have not reproduced the original, but that does not mean the original is real.
Seperate the channels and look at the "Word" output (red channel) and it appears the words have been moved horizontally to line them up with the original fax. Notice that the first letter in each word lines up much better than the last one, and that the spaces seem to be different sizes between the words (compare "Harris gave" to "gave me" for instance). In addition the blue channel is not the same resolution as the pdf file, it appears to have been scaled down by a non-integer, which blurs and thickens the original letters. Also they put red*blue into the green channel to make it look more consistent.
However I have to say that zooming in on the original PDF documents does reveal some strange artifacts. They seem to have been a 3 or 4 level image, with very even gray fringes around the letters. The "th" is also questionable, though there certainly were symbol balls for the Selectric I'm not really certain if it included a small th that looks so exactly scaled. I would also question if anybody would really title the thing "CYA". Because of these doubts it is rather foolish of somebody to make bogus tests when there may be real proof coming, it will only discredit you.
My quick opinion is that if this test is real then these certainly are forgeries, produced just as he said.
Though proportional-spaced typewriters existed then and were quite common (despite claims to the contrary by some people here), they were still mechanical devices. There were only 4 (perhaps 5) letter widths possible, and the numbers (the "en-space") were 3 units (this is for the IBM Selectric I am familiar with). This produces obvious alignement vertically between far more letters than the Word output, as there is a 1/3 chance of a letter aligning with one below it. TrueType fonts have 1000 or more possible widths (I may be thinking of PostScript Type 1 fonts).
It is extremely unlikely that a TrueType font matches the widths used by the Selectric for every letter, too.
Since the documents were covered with dirt and distortions, that would indicate an attempt to make a forgery. No excuse about somebody retyping is going to fly.
PS: I'm going to vote for Kerry anyway. Still somebody is being made a fool here. CBS is really going to lose if these are fake, even retracting their claim is not going to help. The Dems can get out of this easily by saying somebody tried to discredit them with fake documents.
If OpenGL v3 added a higher level API
That would be something like Open Inventor.
What? I'm sure there are legitimate arguments for D3D, but saying the C-API of OpenGL makes it "work worse with other languages" is about the most idiotic thing I have ever seen. It is well known that one of the major reasons for limiting an API to C is to make it callable from multiple languages, ie the exact opposite!
I would be much more convinced with an argument that D3D "rejects old languages and does not compromise the interface in order to support them" which is a legitimate argument.
IBM Selectric typewriters had a single back-quote character, and a single forward quote that was also used as apostrophe. I seem to remember that they also had double back and double forward quote characters, but those may have been missing on the proportional-spaced model that I also remember somewhat. The certainly did not have the neutral double quote from ASCII. Early ASCII teletypes also had a neutral single quote, it was changed on newer systems that added the backquote as well as the lower-case characters.
Selectric typewriters also had 1/2 and some other fractions, a copyright symbol, and some others. They did not have curly braces, less/greater, and many other ASCII symbols. You could also change the ball to a "symbol" ball that had greek (this may not have been possible on the proportional models?)
Such machines existed in the early 70's. My mom used one. I'm going to stop repeating this, because the people claiming the forgeries have made asses of themselves by now.
Proportional-spaced IBM Selectric typewriters (and perhaps other brands) existed in the early 70's, and probably long before. My mom used one, and I played with it trying to make pictures on the paper (the spacing offered the chance to make much fancier graphics, but the machine she was using lacked any way to advance by 1 unit, which limited the ability to place things where wanted.) The machine looked exactly like the fixed-spacing machine she had at home, but I don't believe one could be altered to the other, the character widths were hard-coded.
Before you make an ass out of yourself, you might want to check out IBM selectric typewriters from that time. My mom used one of these. If I remember right, there were 4 or perhaps more different character widths, numbers and "en space" were 3. One frustration to a little proto-nerd kid trying to draw pictures on this was that there was no "move by 1 unit" key.
But 43% of all statistics are useless!
Actually this new system is exactly what you want. The windows have an alpha channel (they are 4-channel off-screen images). These images are handed to the compositing manager, not back to your program, which puts them on the screen.
The only difference from what you want is that the compositing manager can be replaced and can interpret the alpha in other ways besides just stacking the windows. This means that the drop shadows can be added at the compositing step, rather than just making the window bigger with a partially-transparent gray edge. Also means things like fading out the background windows can be done (OS/X obviously redraws the windows to fade the inactive ones).
Running older X11 applications, including ones from the Sun workstation, will work.
The opposite is not true, programs that use these extensions on Linux will not work on remote X servers without them. Now, in theory, programs are supposed to test for the extensions and work around them if they are missing, but I know in reality it is much easier to just assume the extension is there, or to quit with an error if it is not.
In fact the desire to have programs work without the extension has led to much of the stagnation of X. Any complicated extension has been rejected because emulating it is much more work than just not using it at all.
It would help a lot if the authors would produce a library that uses the extension and also provides the emulation if the extension is missing. This was done in Xft and is probably the reason why it was adopted in what seemed like weeks by every piece of Linux software. But it is a lot of work so I don't expect this to be provided for these extensions.
Several of the responders seem confused here. What is wanted is not to literally put every detail of this into the kernel, no more than the details of ext2 are in the base kernel code. Nor does this mean that the entire database interface must be in the kernel.
What is needed is the ability to take any selected object, anywhere, identify it with a "filename" that is a null-terminated string of bytes, and send it to open() and get a link to that object. Then when that SMB selection is made, it can be sent to OpenOffice and it will, guaranteed, open it! The filename can be complete gibberish, OpenOffice does not have to understand it, it just has to rely that it can open it and get the same thing the user expected.
I don't understand why anybody thinks we want to use any interface other than open() to get at objects. It seems painfully obvious to me, yet both Gnome and KDE seem to think we need to learn complex, incompatable interfaces.
It also does not help that there are morons out there pushing "case independence" and "wide characters" and i18n using anything other than utf-8. All of these are a serious impediment to getting this done correctly.
It would help a lot if tar would do it if you just provided -z instead of having to remember to provide -j. Come to think of it, it would be nice if tar just detected compression and you did not have to give it -z either! Can this be done?
The problem was with code that checked if entries in the matrix were zero to avoid degenerate cases that did not work with such rotations. One instance was code to halve a 4x4 perspective transform into two equal transforms. The very small numbers produced math overflows. I did worry about real rotations near 90 producing such small numbers, but it was not possible to produce such small numbers from values of 90 +/- D where D is the minimum delta of a double at a value of 90. Therefore the only way to get it was due to the error between the representable value of M_PI_2 and the perfect value.
Certainly like you say the code needed to be fixed. My post was to point out that the correct fix is to use matricies at a much higher level, abandoning angles as soon as possible. Quaternions are also useful.
Right-angle rotations are betters specified as matricies, where all the coefficients are 1 and 0. I have certainly been burned by math errors where exactly 90 turned into approximatly pi/2 and that turned into 1e-9 in the cos() and this error was later magnified to completely wrong output.
Also use M_PI_2 if you can in your code (of coure Microsoft did not put that in their math.h header file, copy it from the linux header file). I would expect any 3D language to provide a symbol that produces that value with maximum accuracy.
All computer companies say that. It apparently gets them a discount on the Windows bulk pricing.
Quite likely if Linux takes off and they want to buy it from RedHat and RedHat gave them a discount for doing so, they would print "HP recommends RedHat Linux" right next to the Microsoft ad, and look even more silly.