I do not understand this. I use both Gimp and Photoshop, and the way you draw a straight line is EXACTLY THE SAME IN BOTH PROGRAMS! (you hold down shift and click the mouse, Gimp at least shows a preview rubber band, both have the annoying bug that you have to turn off pressure sensitivity if using a pen).
Photoshop does have rectangle and oval buttons. But unless I am spacing out I don't see any other way to paint straight lines, so I do not understand this repetitive complaint about Gimp.
Although there certainly is a difference, you are acting ignorant if you claim more people shut off their Apple computers than let them go to sleep. Visited plenty of houses and if you move the mouse their Apple desktop or laptop wakes up. Never seen one turned off.
The ability to turn on the power on LAN signal certainly requires slightly different hardware than the wake on LAN signal, I have no idea whether or which is common in PC hardware. I doubt Microsoft can magically make power-on-LAN work without the NIC supporting it!
All in all, this seems like something that should have been figured out a decade or more ago. I always had NIC cards that had a "wake on LAN" option but I never saw this do anything, that makes sense now as I understand it was looking for a particular packet. Still this was supported in hardware years and years ago and the fact that nobody thought to make an intelligent front-end until now (not just Microsoft, but Apple and Sun and open source and everybody!) seems strange.
You might want to read some literature not produced by Microsoft. Here is another definition: if a program goes into a loop and does not call the "yield" function, it is cooperative multitasking. If no loop can prevent task switching then it is pre-emptive multitasking.
It has NOTHING to do with "what the programmer thought the call is doing". I certainly DO want to "make the argument that every system that offers system functions is a cooperatively multitasking system" (if you limit it to systems where the system functions actually do task switching).
If somebody from Microsoft said in an article that the DOS was "pre-emtively multitasked" when in fact a loop locks up the computer, they were WRONG. Sorry to break it to you, but the people at Microsoft are not infallable gods. Boo hoo.
Once again, you are making up your own definitions.
"pre-emptive multitasking" means that at an ASYNCHRONOUS point the task may be switched.
"cooperative multitasking" means that the current task executes an instruction to indicate a yield() point.
It does not matter if that instruction is an int21 and that the author of the running program did not know it would do a yield(). It is still cooperative multitasking.
Please try to understand, you are driving people here crazy with your mistake!
Please don't post any more inane reponses until you CLEARLY UNDERSTAND THIS. Read the "It does not matter" paragraph above until you understand what it is saying. Or read any of the hundred other responses trying to explain this to you.
The DOS application "releases the CPU" the moment it does an int21 to make a system call. The int21 causes code that IS NOT PART OF THE PROGRAM to be executed. Microsoft in fact wrote this code, and replaced whatever MSDOS did with new code that did a Windows yield() and then simulated the MSDOS result.
You are the one that is confused. "cooperative" does NOT mean "the program is aware of the multitasking", that is your definition.
Sorry but you have the wrong definition of "cooperative multitasking". You seem to think it means "the program is aware of the multitasking".
What it means is that a call FROM the program is the only point where task switching can happen. As several have pointed out, you can make MSDOS programs cooperatively multitask by making the int21 call that they do all the time to read/write files, and the BIOS interrupt used to read from the keyboard, also do a yield() call.
Preemptive multitasking means that some OUTSIDE, and regular, interrupt causes the task to be switched, no matter what the program is doing. A timer interrupt could do this, for instance.
In fact preemptive multitasking, especially on hardware with no protection, requires the programs to be far more "aware" of it than cooperative does, in that they can never be in a state where the task switch can fail. You actually have your definition pretty much backwards.
Duh. This is not going to make your Edsel engine stop when you approach traffic lights and thus ruin it and cause traffic accidents.
What this is going to do is add to MODERN cars, that have not been built yet. They will include this circuit and starters and engines that are DESIGNED to start/stop rapidly in a way to save gas. This is already done by most hybrids already. And any pressure on the accelerator will override it, this is not big brother trying to stop your constitutional right to drive.
Are you sure that Windows 95 was the first one to really fully use the 32-bit addressing?
I agree that 95 had some major GUI improvements, and one of the few real innovations from Microsoft: they realized that the "icon" (ie the task bar entry) could stay on the screen even when the window was not "iconized" (minimized in their wording). This really was an innovation, everbody else including Mac, all X11 window managers, SGI and NeXT seemed to think the window icon disappeared when the window was not iconized.
Microsoft also removed the divider line between the resize borders and the window contents, which I also consider a major graphical innovation. I was doing that in my hacking on NeXT (bypassing NeXTstep to do so) and was very pleased to see my idea used in a major product (I doubt they copied me...).
He did not describe preemtive multitasking. By "interrupt" he really meant the equivalent of a DOS call to the kernel, which was done with the "int21" instruction, not an async interrupt from a hardware device such as a clock.
I find it hard to believe they made the DOS box preemptive but made Windows programs (where they had a lot more control over what the program was capable of doing) cooperative. Therefore I strongly believe you are wrong and the grandparent is correct.
Relax. In his world, any pollution produced by any company that is subject to *any* law whatsoever (such as income tax...) means that all that pollution is actually caused by the government. Logic does not enter into this discussion.
You are right, I missed the part that some of the work was done before being employed by the company. I was under the impression that he was employed, took some GPL code, and wrote an appliation for internal use, and somehow thought that having used GPL code somehow overrode his employee's ownership of the resulting copyright.
This is hardly GPL specific. If the author gave or sold the copyrights to their previous version to anybody else in any way, the company could not somehow rescind those rights from those others.
Yes it would be a GPL violation if they distributed the program without the source code. But they are not distributing it!
You are trying to make the "virus" false analogy. I myself am free to take a GPL program and write all kinds of derivatives on my computer and then delete the whole mess or leave it on that disk forever. You seem to think that I am somehow violating the GPL by not distributing my result.
I knew this misconception was going to be the very first post!
If the employer is not distributing the program then they are not violating the copyright. Since the GPL is an exception to copyright that allows you to violate it if you follow certain rules, it is entirely meaningless and without force if you are not violating copyright in the first place.
The employee signed a contract that says the employer owns the copyright on their code. This means they cannot distribute the program without violating their employer's copyright, and probably other parts of the contract. The fact that some of the code is based on GPL is irrelevant, there is nothing the employee can do, except maybe try to convince their employer to release the source code to the program under the GPL.
I don't understand how "ribbons" are less discoverable than a menubar. Just click on all the ribbon titles and you see their entire contents, just like if you clicked on each menu title in a menu bar.
The reason for conversion is if the raw data is in a colorspace that is inconvenient to do calculations in. By "inconvenient" I mean that the gamma response is anything other than linear or the sRGB curve (and the sRGB curve is just allowed because of a large number of algorithims designed to work with that space). Color primaries are not as much of a problem, in a linear floating point space it is actually pretty trivial to convert between them.
Due to the pigeonhole principle it is impossible to losslessly convert 12 bit data to 12 bits in a different gamma. You must have more bits so that values that are moved "closer together" can still end up different (there will also be lots of values that are not converted to).
As far as I can tell if you are going to use 16 bits then you might as well use half floats. They process faster, have direct GPU support, they support HDR, and there is no lossage when you multiply by a constant. The log-like floating point representation also matches the gamma response of many devices so that the extra bits are efficiently distributed so that the you can convert to linear without data loss.
Therefore I think any attempt to use more than 8 bits for anything other than floating point is a mistake.
Photoshop is not that useful for moving images, so it is not used for the rig paint out.
Here they run Photoshop under Wine on the Linux machines, so they certainly feel a need to use it. I don't think they have upgraded the install to CS5.
Gimp one looks just like the autoscaling done by Firefox, while the other one looks pretty blurry to me. However Firefox may be using the same algorithm as GIMP and for some reason they both differ from most other programs?
I'm saying that a 12 bit image means you have inserted 3 extra levels between each of the levels of an 8 bit image. Assigning magical properties to this is a mistake. Most of the magical properties people are assuming can be made available by using floating point.
1) How is the floating point image going to have any more detail than what the sensor already captured (which can be contained fully in the 16-bit format)?
If you convert your image to display correctly on the screen then you must have more bits than the hardware produced, otherwise you will have lost data, due to the pigeonhole principle.
2) Which cameras capture floating point data natively?
At the lowest hardware level most are using digital counters. However the conversion software coming with the camera often does use a floating point api. The RED camera is a good example.
Some pointers to actual quotes might make your argument believable.
The only thing I have seen is confusion of gallons and barrels. I have never seen much variation from 5-6K barrels a day, or 200-250K gallons.
One thing I think they should be chastised for is comparing to the Exxon Valdez and quoting that in barrels and this in gallons at the same time. I'm pretty certain I saw that, it would be a good one for you to search for if you want actual proof of your arguments.
According to the article, the *DID* have the domes sitting around. They had been constructed for capping some broken shallow-water wells for Katrina. The dome has to be modified to exactly fit where the pipes and things sticking out of the well are, and also modified for the deeper water. I think if they were building from scratch it would have taken longer.
Slashdot Mirror
I do not understand this. I use both Gimp and Photoshop, and the way you draw a straight line is EXACTLY THE SAME IN BOTH PROGRAMS! (you hold down shift and click the mouse, Gimp at least shows a preview rubber band, both have the annoying bug that you have to turn off pressure sensitivity if using a pen).
Photoshop does have rectangle and oval buttons. But unless I am spacing out I don't see any other way to paint straight lines, so I do not understand this repetitive complaint about Gimp.
Although there certainly is a difference, you are acting ignorant if you claim more people shut off their Apple computers than let them go to sleep. Visited plenty of houses and if you move the mouse their Apple desktop or laptop wakes up. Never seen one turned off.
The ability to turn on the power on LAN signal certainly requires slightly different hardware than the wake on LAN signal, I have no idea whether or which is common in PC hardware. I doubt Microsoft can magically make power-on-LAN work without the NIC supporting it!
All in all, this seems like something that should have been figured out a decade or more ago. I always had NIC cards that had a "wake on LAN" option but I never saw this do anything, that makes sense now as I understand it was looking for a particular packet. Still this was supported in hardware years and years ago and the fact that nobody thought to make an intelligent front-end until now (not just Microsoft, but Apple and Sun and open source and everybody!) seems strange.
Good one. For anybody hiding signatures, the original joke had this:
"93rd rule of Slashdot: No matter how obvious my sarcasm is, my comment will be taken seriously by someone."
Whoosh!
You might want to read some literature not produced by Microsoft. Here is another definition: if a program goes into a loop and does not call the "yield" function, it is cooperative multitasking. If no loop can prevent task switching then it is pre-emptive multitasking.
It has NOTHING to do with "what the programmer thought the call is doing". I certainly DO want to "make the argument that every system that offers system functions is a cooperatively multitasking system" (if you limit it to systems where the system functions actually do task switching).
If somebody from Microsoft said in an article that the DOS was "pre-emtively multitasked" when in fact a loop locks up the computer, they were WRONG. Sorry to break it to you, but the people at Microsoft are not infallable gods. Boo hoo.
Once again, you are making up your own definitions.
"pre-emptive multitasking" means that at an ASYNCHRONOUS point the task may be switched.
"cooperative multitasking" means that the current task executes an instruction to indicate a yield() point.
It does not matter if that instruction is an int21 and that the author of the running program did not know it would do a yield(). It is still cooperative multitasking.
Please try to understand, you are driving people here crazy with your mistake!
Please don't post any more inane reponses until you CLEARLY UNDERSTAND THIS. Read the "It does not matter" paragraph above until you understand what it is saying. Or read any of the hundred other responses trying to explain this to you.
Sorry I was trying to be satirical but it did not come out right.
You still don't seem to get it.
The DOS application "releases the CPU" the moment it does an int21 to make a system call. The int21 causes code that IS NOT PART OF THE PROGRAM to be executed. Microsoft in fact wrote this code, and replaced whatever MSDOS did with new code that did a Windows yield() and then simulated the MSDOS result.
You are the one that is confused. "cooperative" does NOT mean "the program is aware of the multitasking", that is your definition.
Sorry but you have the wrong definition of "cooperative multitasking". You seem to think it means "the program is aware of the multitasking".
What it means is that a call FROM the program is the only point where task switching can happen. As several have pointed out, you can make MSDOS programs cooperatively multitask by making the int21 call that they do all the time to read/write files, and the BIOS interrupt used to read from the keyboard, also do a yield() call.
Preemptive multitasking means that some OUTSIDE, and regular, interrupt causes the task to be switched, no matter what the program is doing. A timer interrupt could do this, for instance.
In fact preemptive multitasking, especially on hardware with no protection, requires the programs to be far more "aware" of it than cooperative does, in that they can never be in a state where the task switch can fail. You actually have your definition pretty much backwards.
Duh. This is not going to make your Edsel engine stop when you approach traffic lights and thus ruin it and cause traffic accidents.
What this is going to do is add to MODERN cars, that have not been built yet. They will include this circuit and starters and engines that are DESIGNED to start/stop rapidly in a way to save gas. This is already done by most hybrids already. And any pressure on the accelerator will override it, this is not big brother trying to stop your constitutional right to drive.
Are you sure that Windows 95 was the first one to really fully use the 32-bit addressing?
I agree that 95 had some major GUI improvements, and one of the few real innovations from Microsoft: they realized that the "icon" (ie the task bar entry) could stay on the screen even when the window was not "iconized" (minimized in their wording). This really was an innovation, everbody else including Mac, all X11 window managers, SGI and NeXT seemed to think the window icon disappeared when the window was not iconized.
Microsoft also removed the divider line between the resize borders and the window contents, which I also consider a major graphical innovation. I was doing that in my hacking on NeXT (bypassing NeXTstep to do so) and was very pleased to see my idea used in a major product (I doubt they copied me...).
He did not describe preemtive multitasking. By "interrupt" he really meant the equivalent of a DOS call to the kernel, which was done with the "int21" instruction, not an async interrupt from a hardware device such as a clock.
I find it hard to believe they made the DOS box preemptive but made Windows programs (where they had a lot more control over what the program was capable of doing) cooperative. Therefore I strongly believe you are wrong and the grandparent is correct.
Relax. In his world, any pollution produced by any company that is subject to *any* law whatsoever (such as income tax...) means that all that pollution is actually caused by the government. Logic does not enter into this discussion.
That's bogus, your reading would disallow internal redistribution of the program, which the FSF has clearly stated is allowed.
You are right, I missed the part that some of the work was done before being employed by the company. I was under the impression that he was employed, took some GPL code, and wrote an appliation for internal use, and somehow thought that having used GPL code somehow overrode his employee's ownership of the resulting copyright.
This is hardly GPL specific. If the author gave or sold the copyrights to their previous version to anybody else in any way, the company could not somehow rescind those rights from those others.
What a load of bullshit.
The breakup contract could have contained a requirement that the spun off company only use the application for internal use and never distribute it.
You are lying and this is a piece of FUD.
Yes it would be a GPL violation if they distributed the program without the source code. But they are not distributing it!
You are trying to make the "virus" false analogy. I myself am free to take a GPL program and write all kinds of derivatives on my computer and then delete the whole mess or leave it on that disk forever. You seem to think that I am somehow violating the GPL by not distributing my result.
I knew this misconception was going to be the very first post!
If the employer is not distributing the program then they are not violating the copyright. Since the GPL is an exception to copyright that allows you to violate it if you follow certain rules, it is entirely meaningless and without force if you are not violating copyright in the first place.
The employee signed a contract that says the employer owns the copyright on their code. This means they cannot distribute the program without violating their employer's copyright, and probably other parts of the contract. The fact that some of the code is based on GPL is irrelevant, there is nothing the employee can do, except maybe try to convince their employer to release the source code to the program under the GPL.
I don't understand how "ribbons" are less discoverable than a menubar. Just click on all the ribbon titles and you see their entire contents, just like if you clicked on each menu title in a menu bar.
The reason for conversion is if the raw data is in a colorspace that is inconvenient to do calculations in. By "inconvenient" I mean that the gamma response is anything other than linear or the sRGB curve (and the sRGB curve is just allowed because of a large number of algorithims designed to work with that space). Color primaries are not as much of a problem, in a linear floating point space it is actually pretty trivial to convert between them.
Due to the pigeonhole principle it is impossible to losslessly convert 12 bit data to 12 bits in a different gamma. You must have more bits so that values that are moved "closer together" can still end up different (there will also be lots of values that are not converted to).
As far as I can tell if you are going to use 16 bits then you might as well use half floats. They process faster, have direct GPU support, they support HDR, and there is no lossage when you multiply by a constant. The log-like floating point representation also matches the gamma response of many devices so that the extra bits are efficiently distributed so that the you can convert to linear without data loss.
Therefore I think any attempt to use more than 8 bits for anything other than floating point is a mistake.
Photoshop is not that useful for moving images, so it is not used for the rig paint out.
Here they run Photoshop under Wine on the Linux machines, so they certainly feel a need to use it. I don't think they have upgraded the install to CS5.
Gimp one looks just like the autoscaling done by Firefox, while the other one looks pretty blurry to me. However Firefox may be using the same algorithm as GIMP and for some reason they both differ from most other programs?
I'm saying that a 12 bit image means you have inserted 3 extra levels between each of the levels of an 8 bit image. Assigning magical properties to this is a mistake. Most of the magical properties people are assuming can be made available by using floating point.
1) How is the floating point image going to have any more detail than what the sensor already captured (which can be contained fully in the 16-bit format)?
If you convert your image to display correctly on the screen then you must have more bits than the hardware produced, otherwise you will have lost data, due to the pigeonhole principle.
2) Which cameras capture floating point data natively?
At the lowest hardware level most are using digital counters. However the conversion software coming with the camera often does use a floating point api. The RED camera is a good example.
Some pointers to actual quotes might make your argument believable.
The only thing I have seen is confusion of gallons and barrels. I have never seen much variation from 5-6K barrels a day, or 200-250K gallons.
One thing I think they should be chastised for is comparing to the Exxon Valdez and quoting that in barrels and this in gallons at the same time. I'm pretty certain I saw that, it would be a good one for you to search for if you want actual proof of your arguments.
According to the article, the *DID* have the domes sitting around. They had been constructed for capping some broken shallow-water wells for Katrina. The dome has to be modified to exactly fit where the pipes and things sticking out of the well are, and also modified for the deeper water. I think if they were building from scratch it would have taken longer.