For the first version, a Microsoft fanboy here has repeatedly posted something about "rewrite the driver to use LDDL" or something like that. Ie perhaps the driver has to be rewritten to a new interface, which is perfectly reasonable, *IF* the same is true of DirectX drivers.
Unfortunatly nobody seems to be backing up this claim with any real information. An 3DLabs spokesman has said this is false, but there is no reason to believe him either, perhaps they want to avoid writing it or want an excuse for some problem with their OpenGL.
They have equally little control over hardare drivers for DirectX. Once the call goes in, it does not matter whether Microsoft invented the call or not. If the driver thinks it should change a pixel on the screen, it will.
So it is true that they have to tell the driver to draw into a given rectangle in memory that may not be on the screen. This is true for both DirectX drivers and OpenGL drivers. Logically there should be a need to rewrite both of them to work correctly with Vista.
Unfortunatly so much shit is being thrown around that I cannot get a clear answer.
If in fact DirectX drivers work with no changes but OpenGL does not, or that the changes needed for DirectX drivers cannot work for OpenGL, then I would believe the worst possible theory about why Microsoft is doing this. However I find it hard to believe that actual software writers, even at microsoft, would do this with a clear conciense, surely they should be unable to sleep at night knowing exactly what they are doing to everybody. So I am going to have to withhold judgement until we find out what really is going on.
Sorry what I wanted to point out that was "awfully strange" about the hardware OpenGL breaking Vista's desktop compositing:
1. There is no word that DirectX hardware drivers break things in the same way. But there is absolutely no reason why this would be true for one type of driver and not the other. Win XP DirectX drivers draw on the screen buffer just like OpenGL drivers do.
2. It would seem that keeping the old XP interface around and making it work with new programs and correctly window the drivers output is VERY difficult. If there is really a technical problem I would expect the solution to be more like the hardware driver is completely ignored, rather than the display changing, or that the driver fails completely and leaves empty windows or draws over the screen. Sounds to me more like a purposeful design to make things look bad.
For these two reasons I feel the story is either:
1. An actual case of Microsoft being assholes and screwing everybody in their mad attempt to disable any competition.
2. A big anti-Microsoft FUD.
Unfortunatly I don't see any actual logical technical explanation for this.
There seems to be a lot of (mis)information being flung around. My current understanding is this:
By default the OpenGL library provided by Microsoft will match OpenGL1.4 and will be emulated atop DirectX, which has been modified to work in windows rather than take over the screen. I doubt this is much worse than the OpenGL emulation Microsoft provides now (converse claims that this will make OpenGL work better are false, the OpenGL emulation works on all cards that support DirectX now, and calls some of the DirectX functions).
Graphics cards that have OpenGL drivers work by pretty much replacing the entire OpenGL implementation. The claim is that if you use such a driver the desktop compositing will stop working, making Vista look like XP. This seems awfully strange for two reasons: first all OpenGL implementations can draw to offscreen buffers that could then be composited, and more importantly there is no indication of similar breakage for "old" DirectX drivers.
There is a claim that something called "LMML" or something like that is a new interface that OpenGL drivers must conform to and then they will work with Vista. This sounds perfectly reasonable. However it is impossible to confirm anything about this, the Opengl.org posters and the sketch at Siggraph acted as though such a thing absolutley does not exist. It is also possible that what exists has been designed such that fast or completel OpenGL is impossible. I would appreticate any actual information.
WINE tries to be compatable with the *newest* DirectX. Not with some old version. Any additions Microsoft makes to DirectX are implemented in WINE as fast as possible.
Also it is absolutly SLOW. OpenGL Windows games are able to run under WINE at twice or more the frame rates as DirectX ones. Apparently the overhead is easily 50%, and this is from people who don't have an incentive to make it perform badly.
Furthermore, venders can create LMMH compatible OpenGL drivers
Can you go into detail on what this is? After reading all the articles and froth from both sides, this is the first time this has been mentioned.
It sounds a little like "Vista is incompatable with *old* OpenGL drivers", which is perfectly logical (though a bit evil if it is for some reason compatable with old DirectX drivers...). If the vendors just have to rewrite their drivers and get full speed, preferably with a scheme so that the DirectX and OpenGL paths can be mixed to reuse code or the emulation be reversed so that DirectX is emulated using the OpenGL functions, than this would be perfectly reasonable.
Unfortunatlye NOTHING I have heard indicates this. The claim was that hardware-accelerated OpenGL in desktop windows was impossible unless Microsoft changes things.
I actually made about $3000 off that, due to getting one of the IPO shares. Could have made about twice that if I had just sold as soon as possible, rather than wait a month.
That stock was certainly inflated to a hundred times what it was worth, apparently because of the "Linux" buzzword. Actually a very good indication of how nuts everything was. Anybody with the slightest knowledge of how Linux worked would know that if Linux took off, the money would go to a hardware manufacturer (like Dell or IBM) who could actually sell something.
That would be X10 in 1984, not X1. I think X was started several years earlier, but about the same time as the Lisa, which I did not realize was that early (I saw working X and a very preliminary NeWS implementation before I saw a Lisa, but that was at least a year after the Lisa was released).
You are probably right that it was pre-X, but I think that everybody was convinced overlapping windows were the way to go, based on negative experiences with the Andrew system. The Apollo system also had overlapping windows, I think.
No you misunderstood. The voter does not keep the paper, they put it in a ballot box. Keeping a proof of how you voted would allow vote buying (ie your boss insists you show you voted for his candidate or your are fired).
I think having the voter move the paper from the machine to the box would inspire more confidence that it really is their vote (the machine *may* be incinerating the enclosed paper you see and printing another...) However I can see problems with some scheme where many voters are somehow made to submit bogus receipts in place of their real ones, screwing up an election by invalidating many results due to mismatches between the paper and electronic votes. This may be why the "paper behind glass" schemes are being proposed.
The "trash to eject" feature (or misfeature, if you insist) was not in the original Lisa or Mac. It was added later. So the original design did not mangle the metaphor.
The main problem is that the only quick operation was apple-E to eject the disk, which did not forget about the contents. This caused it to (due to various other bugs and misfeatures) to randomly ask for you to re-insert some disk you ejected long ago. The official method of "eject and forget" was difficult (I can't even remember what it was) so they added the drag-to-trash method so people had a quick way to do what they really wanted (not sure why they could not just change apple-E to do it, however...)
I thought Windows was purposly designed so that you removed media by "ripping it out". They do this by flushing the data to the device as often as possible, ignoring errors, and recovering cleanly if the device was miswritten. Sounds bad, but the chances of failure are miniscule (much more likely to mess up the connector plugging it in), so IMH this is *right* and one of the areas where Windows is well ahead of either Linux or OS/X. Micorsoft does "get it" sometimes...
Although I agree about the menus and trash, overlapping windows was well established in existing X servers at that time. X's main difference from the W system used by Andrew was that the windows were not tiled but instead overlapping.
The Xerox machine, as well as Lisp machines (which nobody here seems to be mentioning) used tiled windows. The most obvious difference with even the first versions of X is overlapping windows.
An open-source windows would be very quickly modified to be compatable with Linux/POSIX/Unix (ie to the point that non-GUI Linux apps run with no changes, and all scripts and shells work).
It is hard to say if Linux would be modified as well, probably to pick up gui and services stuff from Windows. If this starts to happen the result is likely to be a merge.
Um, I would think that this "family values" stuff is a shift to the *right*, not left.
Otherwise I agree that this is a stupid move. She seems to be picking up exactly the least-liked part of the "right" here. There is a significant number of people who are libartarian (even if they don't say they are). They like some of the "right", and some of the "left". But what Hillary is starting to talk about is definatly the *disliked* part of the "right". I'm sure it is going to result in a disaster and we will have Bush's brother in office next...
In Illustrator, for example, you never need to read the keyboard chart because you are informed of the keyboard shortcut for any command when you mouse over it.
Please explain how you get a tooltip that shows that holding down space will pan the image.
I agree with the above, having worked on this myself. It is quite possible to write a program using the Quartz drawing interface in just C (not even C++).
The toolkit (Cocoa) requires Objective-C. But it also means you are not writing a portable application (unless you intend to emulate Cocoa on the other platforms), so that is completely out of the question. Any kind of portable program can also easily avoid Objective-C.
Just searching around for info on this quickly led to what sounds like a much larger case that went absolutly the other way. Possibly the article you quoted is being somewhat selective, it did quickly mention that the defendant did not show up in court, which hurt his case:
Trinity Western University From Wikipedia, the free encyclopedia.
Trinity Western University is a private Christian liberal arts and sciences university located in Langley, British Columbia, Canada, enrolling approximately 2,850 students.
Most notable about the school is its requirement that all students sign an agreement that prohibits, among other things, premarital sex, and homosexual behaviour.
In 1995, TWU attempted to certify students going through its teaching program, but the British Columbia College of Teachers denied accreditation of TWU's program, arguing that the agreement students must sign is discriminatory and that those graduating from Trinity Western's program might discriminate against gay students. The lower courts in British Columbia and, later, the Supreme Court of Canada, ruled in favour of Trinity Western, stating that there was no basis for the BCCT's decision.
Judging by the posts (filtered at +2) it looks like a landslide victory for Microsoft's side. The guy broke a contract.
If you want to be cynical, possibly the reason is that lots of Slashdot readership is under such contracts as well. It is unlikely that this guy winning is going to change anything for the average IT worker. So they don't want to see some rich and famous guy get away with something they can't do themselves.
Although this is pretty neat, I think far more useful would be a new form of "drag and drop" which just happens to be compatable with the the X middle mouse actions, but works for all objects: Select an object (in a program-specific way) to start a "drag", and click the middle mouse button to do a "drop". Normal interface is also supported where you push down on a selected object and then let go at the drop location. Pushing down the middle mouse button and holding it will result in exactly the same state as though you had dragged to that point, with the same cursor feedback, etc.
This has a whole lot of advantages: 1. You can drop on anything, including windows you have not created yet. As long as you can get to a window setup without selecting a draggable object (and thus starting a new drag) you can drop on it. 2. You can also drop the same source more than once, if that has a meaningful result. 3. It is obvious and intuitive how to abort a "drag and drop".
You explanation makes it sound very safe, but I have my doubts your explanation is correct.
It appears GM added commands to use the contents of files for it's own purposes, and the bug is that other pages not controlled by GM can get at these commands.
I find it hard to believe they implemented these commands to check for the files to be world-readable. First this would mean that anybody using GM for it's intended purpose would have to make the file it is reading world-readable, which would be sort of a security problem in it's own right (say it has secret information in it that you don't want other users of the computer to see).
Second it requires quite a bit of annoying code to check for world-readable, as opposed to just trying to read the file, and you are implying they did this correctly both for Windows and Linux.
I'm just not buying it, without a more compelling explanation as to why they would have implemented it this way.
Slashdot Mirror
The claim is that latter.
For the first version, a Microsoft fanboy here has repeatedly posted something about "rewrite the driver to use LDDL" or something like that. Ie perhaps the driver has to be rewritten to a new interface, which is perfectly reasonable, *IF* the same is true of DirectX drivers.
Unfortunatly nobody seems to be backing up this claim with any real information. An 3DLabs spokesman has said this is false, but there is no reason to believe him either, perhaps they want to avoid writing it or want an excuse for some problem with their OpenGL.
They have equally little control over hardare drivers for DirectX. Once the call goes in, it does not matter whether Microsoft invented the call or not. If the driver thinks it should change a pixel on the screen, it will.
So it is true that they have to tell the driver to draw into a given rectangle in memory that may not be on the screen. This is true for both DirectX drivers and OpenGL drivers. Logically there should be a need to rewrite both of them to work correctly with Vista.
Unfortunatly so much shit is being thrown around that I cannot get a clear answer.
If in fact DirectX drivers work with no changes but OpenGL does not, or that the changes needed for DirectX drivers cannot work for OpenGL, then I would believe the worst possible theory about why Microsoft is doing this. However I find it hard to believe that actual software writers, even at microsoft, would do this with a clear conciense, surely they should be unable to sleep at night knowing exactly what they are doing to everybody. So I am going to have to withhold judgement until we find out what really is going on.
Sorry what I wanted to point out that was "awfully strange" about the hardware OpenGL breaking Vista's desktop compositing:
1. There is no word that DirectX hardware drivers break things in the same way. But there is absolutely no reason why this would be true for one type of driver and not the other. Win XP DirectX drivers draw on the screen buffer just like OpenGL drivers do.
2. It would seem that keeping the old XP interface around and making it work with new programs and correctly window the drivers output is VERY difficult. If there is really a technical problem I would expect the solution to be more like the hardware driver is completely ignored, rather than the display changing, or that the driver fails completely and leaves empty windows or draws over the screen. Sounds to me more like a purposeful design to make things look bad.
For these two reasons I feel the story is either:
1. An actual case of Microsoft being assholes and screwing everybody in their mad attempt to disable any competition.
2. A big anti-Microsoft FUD.
Unfortunatly I don't see any actual logical technical explanation for this.
There seems to be a lot of (mis)information being flung around. My current understanding is this:
By default the OpenGL library provided by Microsoft will match OpenGL1.4 and will be emulated atop DirectX, which has been modified to work in windows rather than take over the screen. I doubt this is much worse than the OpenGL emulation Microsoft provides now (converse claims that this will make OpenGL work better are false, the OpenGL emulation works on all cards that support DirectX now, and calls some of the DirectX functions).
Graphics cards that have OpenGL drivers work by pretty much replacing the entire OpenGL implementation. The claim is that if you use such a driver the desktop compositing will stop working, making Vista look like XP. This seems awfully strange for two reasons: first all OpenGL implementations can draw to offscreen buffers that could then be composited, and more importantly there is no indication of similar breakage for "old" DirectX drivers.
There is a claim that something called "LMML" or something like that is a new interface that OpenGL drivers must conform to and then they will work with Vista. This sounds perfectly reasonable. However it is impossible to confirm anything about this, the Opengl.org posters and the sketch at Siggraph acted as though such a thing absolutley does not exist. It is also possible that what exists has been designed such that fast or completel OpenGL is impossible. I would appreticate any actual information.
FIFTY PERCENT of our sales are Linux. This is for a $7999/seat graphics software.
WINE tries to be compatable with the *newest* DirectX. Not with some old version. Any additions Microsoft makes to DirectX are implemented in WINE as fast as possible.
Also it is absolutly SLOW. OpenGL Windows games are able to run under WINE at twice or more the frame rates as DirectX ones. Apparently the overhead is easily 50%, and this is from people who don't have an incentive to make it perform badly.
All advanced 3D graphics applications (like Maya) use OpenGL and use windows.
Games are not everything.
This is in fact an attempt to kill off portability to Linux/Mac of advanced graphics applications.
Furthermore, venders can create LMMH compatible OpenGL drivers
Can you go into detail on what this is? After reading all the articles and froth from both sides, this is the first time this has been mentioned.
It sounds a little like "Vista is incompatable with *old* OpenGL drivers", which is perfectly logical (though a bit evil if it is for some reason compatable with old DirectX drivers...). If the vendors just have to rewrite their drivers and get full speed, preferably with a scheme so that the DirectX and OpenGL paths can be mixed to reuse code or the emulation be reversed so that DirectX is emulated using the OpenGL functions, than this would be perfectly reasonable.
Unfortunatlye NOTHING I have heard indicates this. The claim was that hardware-accelerated OpenGL in desktop windows was impossible unless Microsoft changes things.
So what exactly is "LMMH"?
I actually made about $3000 off that, due to getting one of the IPO shares. Could have made about twice that if I had just sold as soon as possible, rather than wait a month.
That stock was certainly inflated to a hundred times what it was worth, apparently because of the "Linux" buzzword. Actually a very good indication of how nuts everything was. Anybody with the slightest knowledge of how Linux worked would know that if Linux took off, the money would go to a hardware manufacturer (like Dell or IBM) who could actually sell something.
Microsoft decided to kill off Netscape long before Java existed.
That would be X10 in 1984, not X1. I think X was started several years earlier, but about the same time as the Lisa, which I did not realize was that early (I saw working X and a very preliminary NeWS implementation before I saw a Lisa, but that was at least a year after the Lisa was released).
You are probably right that it was pre-X, but I think that everybody was convinced overlapping windows were the way to go, based on negative experiences with the Andrew system. The Apollo system also had overlapping windows, I think.
No you misunderstood. The voter does not keep the paper, they put it in a ballot box. Keeping a proof of how you voted would allow vote buying (ie your boss insists you show you voted for his candidate or your are fired).
I think having the voter move the paper from the machine to the box would inspire more confidence that it really is their vote (the machine *may* be incinerating the enclosed paper you see and printing another...) However I can see problems with some scheme where many voters are somehow made to submit bogus receipts in place of their real ones, screwing up an election by invalidating many results due to mismatches between the paper and electronic votes. This may be why the "paper behind glass" schemes are being proposed.
The "trash to eject" feature (or misfeature, if you insist) was not in the original Lisa or Mac. It was added later. So the original design did not mangle the metaphor.
The main problem is that the only quick operation was apple-E to eject the disk, which did not forget about the contents. This caused it to (due to various other bugs and misfeatures) to randomly ask for you to re-insert some disk you ejected long ago. The official method of "eject and forget" was difficult (I can't even remember what it was) so they added the drag-to-trash method so people had a quick way to do what they really wanted (not sure why they could not just change apple-E to do it, however...)
I thought Windows was purposly designed so that you removed media by "ripping it out". They do this by flushing the data to the device as often as possible, ignoring errors, and recovering cleanly if the device was miswritten. Sounds bad, but the chances of failure are miniscule (much more likely to mess up the connector plugging it in), so IMH this is *right* and one of the areas where Windows is well ahead of either Linux or OS/X. Micorsoft does "get it" sometimes...
Although I agree about the menus and trash, overlapping windows was well established in existing X servers at that time. X's main difference from the W system used by Andrew was that the windows were not tiled but instead overlapping.
The Xerox machine, as well as Lisp machines (which nobody here seems to be mentioning) used tiled windows. The most obvious difference with even the first versions of X is overlapping windows.
An open-source windows would be very quickly modified to be compatable with Linux/POSIX/Unix (ie to the point that non-GUI Linux apps run with no changes, and all scripts and shells work).
It is hard to say if Linux would be modified as well, probably to pick up gui and services stuff from Windows. If this starts to happen the result is likely to be a merge.
Pixar's Renderman is often used, especially on a renderfarm, even if the modelling is done in Maya or Softimage.
Um, I would think that this "family values" stuff is a shift to the *right*, not left.
Otherwise I agree that this is a stupid move. She seems to be picking up exactly the least-liked part of the "right" here. There is a significant number of people who are libartarian (even if they don't say they are). They like some of the "right", and some of the "left". But what Hillary is starting to talk about is definatly the *disliked* part of the "right". I'm sure it is going to result in a disaster and we will have Bush's brother in office next...
In Illustrator, for example, you never need to read the keyboard chart because you are informed of the keyboard shortcut for any command when you mouse over it.
Please explain how you get a tooltip that shows that holding down space will pan the image.
All modern tablets have a method to emulate middle or right click on the pen.
I agree with the above, having worked on this myself. It is quite possible to write a program using the Quartz drawing interface in just C (not even C++).
The toolkit (Cocoa) requires Objective-C. But it also means you are not writing a portable application (unless you intend to emulate Cocoa on the other platforms), so that is completely out of the question. Any kind of portable program can also easily avoid Objective-C.
Just searching around for info on this quickly led to what sounds like a much larger case that went absolutly the other way. Possibly the article you quoted is being somewhat selective, it did quickly mention that the defendant did not show up in court, which hurt his case:
Trinity Western University
From Wikipedia, the free encyclopedia.
Trinity Western University is a private Christian liberal arts and sciences university located in Langley, British Columbia, Canada, enrolling approximately 2,850 students.
Most notable about the school is its requirement that all students sign an agreement that prohibits, among other things, premarital sex, and homosexual behaviour.
In 1995, TWU attempted to certify students going through its teaching program, but the British Columbia College of Teachers denied accreditation of TWU's program, arguing that the agreement students must sign is discriminatory and that those graduating from Trinity Western's program might discriminate against gay students. The lower courts in British Columbia and, later, the Supreme Court of Canada, ruled in favour of Trinity Western, stating that there was no basis for the BCCT's decision.
Judging by the posts (filtered at +2) it looks like a landslide victory for Microsoft's side. The guy broke a contract.
If you want to be cynical, possibly the reason is that lots of Slashdot readership is under such contracts as well. It is unlikely that this guy winning is going to change anything for the average IT worker. So they don't want to see some rich and famous guy get away with something they can't do themselves.
Although this is pretty neat, I think far more useful would be a new form of "drag and drop" which just happens to be compatable with the the X middle mouse actions, but works for all objects: Select an object (in a program-specific way) to start a "drag", and click the middle mouse button to do a "drop". Normal interface is also supported where you push down on a selected object and then let go at the drop location. Pushing down the middle mouse button and holding it will result in exactly the same state as though you had dragged to that point, with the same cursor feedback, etc.
This has a whole lot of advantages: 1. You can drop on anything, including windows you have not created yet. As long as you can get to a window setup without selecting a draggable object (and thus starting a new drag) you can drop on it. 2. You can also drop the same source more than once, if that has a meaningful result. 3. It is obvious and intuitive how to abort a "drag and drop".
You explanation makes it sound very safe, but I have my doubts your explanation is correct.
It appears GM added commands to use the contents of files for it's own purposes, and the bug is that other pages not controlled by GM can get at these commands.
I find it hard to believe they implemented these commands to check for the files to be world-readable. First this would mean that anybody using GM for it's intended purpose would have to make the file it is reading world-readable, which would be sort of a security problem in it's own right (say it has secret information in it that you don't want other users of the computer to see).
Second it requires quite a bit of annoying code to check for world-readable, as opposed to just trying to read the file, and you are implying they did this correctly both for Windows and Linux.
I'm just not buying it, without a more compelling explanation as to why they would have implemented it this way.