They rewrote a fair amount of the code of the open software they use, or added much to it. That's why the whining from the open software crowd that Apple doesn't do it the way they want.
Links? Or are you just making general assumptions based on one project (KHTML) which they fixed two years ago.
Quote:
Originally Posted by melgross
Apple added many features to FreeBSD and other projects it contributed code to.
Perhaps I've named it wrongly. I was talking about the x86 compatability box. What was it called? I obviously don't remember.
Presumably you are talking about Yellow Box for Windows - that would have allowed developers to write Cocoa apps and compile them as "fat" or "universal" binaries that would run on either PPC Mac OS X or Windows with Yellow Box for Windows installed. That was originally planned and subsequently dropped. However, it is unclear whether it was dropped because there were too many problems getting it to work, or if it was more of a business decision (people were worried that Yellow Box for Windows would reduce the incentive for Windows users to switch to Mac).
You could also be thinking of Red Box. Red Box was/is a myth - a proposed Windows API for OS X - i.e. the equivalent of WINE/Crossover. Apple were never working on a Red Box.
Then, as a developer, I'd be very curious how to activate it. You can't.
You mean "Enable Quartz 2D Extreme" in the Quartz Debug application doesn't actually do anything?
I never played with it myself, but in the early days of 10.4 I read plenty of posts about people's experience with Q2DE, those people having enabled it with Quartz Debug. I had no reason to doubt the veracity of the statements I was reading.
Then, as a developer, I'd be very curious how to activate it. You can't.
I just got hold of Xbench, and ran it on my 1.83 GHz Core Duo MacBook and my girlfriend's 12" 1.5 GHz PB G4, both with and without Q2DE enabled.
On the MacBook, running 10.4.9, it didn't make any difference at all. This isn't too suprising given how Q2DE works and that the MacBook has an integrated GPU - the main aim of Q2DE is to exploit the much faster GPU dedicated VRAM, and minimise the amount of data sent over the slower system bus to the GPU. The MacBook doesn't have dedicated VRAM, so there's no advantage to be had.
However, the 12" PB G4 (running 10.4.8) does have dedicated VRAM, and the difference in performance with Q2DE on is significant. Xbench has 5 Quartz tests: Line, Rectangle, Circle, Bezier and Text.
Results were as follows:
Q2DE off:
Line: 33.92
Rectangle: 37.94
Circle: 37.54
Bezier: 42.80
Text: 74.58
Q2DE on:
Line: 70.70
Rectangle: 49.77
Circle: 57.40
Bezier: 52.54
Text: 29.74
So, the first four tests are faster with it on, Line and Circle especially so. Unfortunately, as I mentioned in an earlier post, text performance falls significantly.
Xbench also has a User Interface Test, and here the situation is even worse:
You mean "Enable Quartz 2D Extreme" in the Quartz Debug application doesn't actually do anything?
I never played with it myself, but in the early days of 10.4 I read plenty of posts about people's experience with Q2DE, those people having enabled it with Quartz Debug. I had no reason to doubt the veracity of the statements I was reading.
Like I said earlier, this was possible in early versions of 10.4. I'm not sure which version disabled it (10.4.3?).
Just look at Quartz Debug's window list. The only ones in green (and therefore Q2DE-accelerated) are Rosetta ones, and those happen to have lots of drawing glitches.
I know it's not the most accurate benchmarking application in the world, but are you really denying that the results I've posted demonstrate that, at least on a PPC mac running 10.4.8, the Quartz Debug application really can turn on Quartz 2D extreme?
I know it's not the most accurate benchmarking application in the world, but are you really denying that the results I've posted demonstrate that, at least on a PPC mac running 10.4.8, the Quartz Debug application really can turn on Quartz 2D extreme?
I've already said earlier that it might work on PowerPC. Feel free to prove that it can me made to work on a MacBook Pro, though.
[QUOTE=JLL;1072935]Links? Or are you just making general assumptions based on one project (KHTML) which they fixed two years ago.
to just give a good example of their innovation which has been given back, and which is considered to be a serious advance: launchd
And no, I'm not going to get into a programming debate about the exact meaning or value of things such as this, because, quite frankly, I have no interest in doing so. I haven programmed in about ten years, so while I do understand a fair amount of what is going on today, I'm too rustry to argue the esoterics anymore.
Presumably you are talking about Yellow Box for Windows - that would have allowed developers to write Cocoa apps and compile them as "fat" or "universal" binaries that would run on either PPC Mac OS X or Windows with Yellow Box for Windows installed. That was originally planned and subsequently dropped. However, it is unclear whether it was dropped because there were too many problems getting it to work, or if it was more of a business decision (people were worried that Yellow Box for Windows would reduce the incentive for Windows users to switch to Mac).
I've already said earlier that it might work on PowerPC. Feel free to prove that it can me made to work on a MacBook Pro, though.
Apple obviously abandoned it for 10.4 after a while. Possibly they realized the weren't going to get it to be useful at that time, and didn't even bother to add it to the Intel version of the OS. Since they wanted to synchronize the two versions, they simply dropped it.
If it does reappear as part of 10.5 that will be good, because it is really necessary to move the OS forwards.
I sense there is a kernal of truth in the points being made. Apple could well be spread too thin for the long haul. We shall see if Leopard is delayed again, and if the iPhone ships as it's supposed to.
Lets be really clear, I know the guy who wrote a large chunk of freebsd. The only company contributing to it much any more is Yahoo, who still has a large effort to contribute back.
The truth with freebsd/openbsd/netbsd is that Apple's code is largely unchanged from the kernel, people always say it's "rewritten by apple", but really the didn't do much to the kernel.
Apple isn't spread too thin, it's really just an issue of do we want to hire somebody and put them straight away on Mac OS X? Is that who you want your OS programmed by? Typically from the people I know working on OS X, they took 3-5 years to be allowed onto the OS X development team. Apple is constantly hiring new programmers, and is enlarging it's campus for another 4000+ employees as of last year. Even with that you're not going to see people walk in the door and fill the spot on a important project, unless someone knows them personally.
The leopard team didn't work on the iPhone, they worked ON THE iPhone OS, which "gasp" is Leopard!!! Exactly what they SHOULD have been working on!!!!! Especially since the product is being released first.
Links? Or are you just making general assumptions based on one project (KHTML) which they fixed two years ago.
That's not rewriting - that's the point of OSS.
ObjC. The main bulk of ObjC work done for GCC is from Apple. The debates among gcc maintainers and Apple are quite extensive and sometimes worthy of a popcorn parking and just watching the whining go back and forth.
GNUstep getting ObjC++ is a complete contribution from Apple, along with a complete rewrite of bison for ObjC/ObjC++.
Links? Or are you just making general assumptions based on one project (KHTML) which they fixed two years ago.
to just give a good example of their innovation which has been given back, and which is considered to be a serious advance: launchd
And no, I'm not going to get into a programming debate about the exact meaning or value of things such as this, because, quite frankly, I have no interest in doing so. I haven programmed in about ten years, so while I do understand a fair amount of what is going on today, I'm too rustry to argue the esoterics anymore.
Lets be really clear, I know the guy who wrote a large chunk of freebsd. The only company contributing to it much any more is Yahoo, who still has a large effort to contribute back.
The truth with freebsd/openbsd/netbsd is that Apple's code is largely unchanged from the kernel, people always say it's "rewritten by apple", but really the didn't do much to the kernel.
Pretty much left FreeBSD's kernel alone and replaced most of it with Mach is what they did. Split it at the upper/lower demarkation and only kept the FreeBSD upper half. That's why Apple changes to the kernel mostly don't make it back to FreeBSD, they are meaningless without the lower Mach portion. Especially since the lock granularity changes have been implemented.
I think Apple has been playing nice with the user-land stuff since it's kernel agnostic.
Wow, glass screens on a phone. That's a brilliant "reliable source" you've got. Did he also tell you that the next MBPro would be made of balsa wood to lower the weight?
Comments
They rewrote a fair amount of the code of the open software they use, or added much to it. That's why the whining from the open software crowd that Apple doesn't do it the way they want.
Links? Or are you just making general assumptions based on one project (KHTML) which they fixed two years ago.
Apple added many features to FreeBSD and other projects it contributed code to.
That's not rewriting - that's the point of OSS.
they left it there for developers to play with until they could.
Then, as a developer, I'd be very curious how to activate it. You can't.
Perhaps I've named it wrongly. I was talking about the x86 compatability box. What was it called? I obviously don't remember.
Presumably you are talking about Yellow Box for Windows - that would have allowed developers to write Cocoa apps and compile them as "fat" or "universal" binaries that would run on either PPC Mac OS X or Windows with Yellow Box for Windows installed. That was originally planned and subsequently dropped. However, it is unclear whether it was dropped because there were too many problems getting it to work, or if it was more of a business decision (people were worried that Yellow Box for Windows would reduce the incentive for Windows users to switch to Mac).
You could also be thinking of Red Box. Red Box was/is a myth - a proposed Windows API for OS X - i.e. the equivalent of WINE/Crossover. Apple were never working on a Red Box.
Then, as a developer, I'd be very curious how to activate it. You can't.
You mean "Enable Quartz 2D Extreme" in the Quartz Debug application doesn't actually do anything?
I never played with it myself, but in the early days of 10.4 I read plenty of posts about people's experience with Q2DE, those people having enabled it with Quartz Debug. I had no reason to doubt the veracity of the statements I was reading.
Then, as a developer, I'd be very curious how to activate it. You can't.
I just got hold of Xbench, and ran it on my 1.83 GHz Core Duo MacBook and my girlfriend's 12" 1.5 GHz PB G4, both with and without Q2DE enabled.
On the MacBook, running 10.4.9, it didn't make any difference at all. This isn't too suprising given how Q2DE works and that the MacBook has an integrated GPU - the main aim of Q2DE is to exploit the much faster GPU dedicated VRAM, and minimise the amount of data sent over the slower system bus to the GPU. The MacBook doesn't have dedicated VRAM, so there's no advantage to be had.
However, the 12" PB G4 (running 10.4.8) does have dedicated VRAM, and the difference in performance with Q2DE on is significant. Xbench has 5 Quartz tests: Line, Rectangle, Circle, Bezier and Text.
Results were as follows:
Q2DE off:
Line: 33.92
Rectangle: 37.94
Circle: 37.54
Bezier: 42.80
Text: 74.58
Q2DE on:
Line: 70.70
Rectangle: 49.77
Circle: 57.40
Bezier: 52.54
Text: 29.74
So, the first four tests are faster with it on, Line and Circle especially so. Unfortunately, as I mentioned in an earlier post, text performance falls significantly.
Xbench also has a User Interface Test, and here the situation is even worse:
Q2DE off:
UI score: 22.17
Q2DE on:
UI score: 2.58
You mean "Enable Quartz 2D Extreme" in the Quartz Debug application doesn't actually do anything?
I never played with it myself, but in the early days of 10.4 I read plenty of posts about people's experience with Q2DE, those people having enabled it with Quartz Debug. I had no reason to doubt the veracity of the statements I was reading.
Like I said earlier, this was possible in early versions of 10.4. I'm not sure which version disabled it (10.4.3?).
Just look at Quartz Debug's window list. The only ones in green (and therefore Q2DE-accelerated) are Rosetta ones, and those happen to have lots of drawing glitches.
I just got hold of Xbench
Don't waste your time with Xbench.
Don't waste your time with Xbench.
I know it's not the most accurate benchmarking application in the world, but are you really denying that the results I've posted demonstrate that, at least on a PPC mac running 10.4.8, the Quartz Debug application really can turn on Quartz 2D extreme?
I know it's not the most accurate benchmarking application in the world, but are you really denying that the results I've posted demonstrate that, at least on a PPC mac running 10.4.8, the Quartz Debug application really can turn on Quartz 2D extreme?
I've already said earlier that it might work on PowerPC. Feel free to prove that it can me made to work on a MacBook Pro, though.
to just give a good example of their innovation which has been given back, and which is considered to be a serious advance: launchd
And no, I'm not going to get into a programming debate about the exact meaning or value of things such as this, because, quite frankly, I have no interest in doing so. I haven programmed in about ten years, so while I do understand a fair amount of what is going on today, I'm too rustry to argue the esoterics anymore.
But here's Apple's developer page on it:
http://developer.apple.com/macosx/launchd.html
Here is the Google page for links, if you are interested, and if the page carries through:
http://www.google.com/search?client=...UTF-8&oe=UTF-8
That's not rewriting - that's the point of OSS.
Some of it is rewriting, and some of it is adding to.
Then, as a developer, I'd be very curious how to activate it. You can't.
I'm late to this thread today, so I see you have already been answered.
Presumably you are talking about Yellow Box for Windows - that would have allowed developers to write Cocoa apps and compile them as "fat" or "universal" binaries that would run on either PPC Mac OS X or Windows with Yellow Box for Windows installed. That was originally planned and subsequently dropped. However, it is unclear whether it was dropped because there were too many problems getting it to work, or if it was more of a business decision (people were worried that Yellow Box for Windows would reduce the incentive for Windows users to switch to Mac).
Yes, exactly. I guess my memory wasn't that bad.
I've already said earlier that it might work on PowerPC. Feel free to prove that it can me made to work on a MacBook Pro, though.
Apple obviously abandoned it for 10.4 after a while. Possibly they realized the weren't going to get it to be useful at that time, and didn't even bother to add it to the Intel version of the OS. Since they wanted to synchronize the two versions, they simply dropped it.
If it does reappear as part of 10.5 that will be good, because it is really necessary to move the OS forwards.
FWIW, John Martellaro seems to agree with me here: http://www.macobserver.com/columns/h.../04/16.1.shtml
In this article by Dennis Sellers, COO Tim Cook apparently feels a need to knock down Martellaro's observations...
I sense there is a kernal of truth in the points being made. Apple could well be spread too thin for the long haul. We shall see if Leopard is delayed again, and if the iPhone ships as it's supposed to.
The truth with freebsd/openbsd/netbsd is that Apple's code is largely unchanged from the kernel, people always say it's "rewritten by apple", but really the didn't do much to the kernel.
The leopard team didn't work on the iPhone, they worked ON THE iPhone OS, which "gasp" is Leopard!!! Exactly what they SHOULD have been working on!!!!! Especially since the product is being released first.
Links? Or are you just making general assumptions based on one project (KHTML) which they fixed two years ago.
That's not rewriting - that's the point of OSS.
ObjC. The main bulk of ObjC work done for GCC is from Apple. The debates among gcc maintainers and Apple are quite extensive and sometimes worthy of a popcorn parking and just watching the whining go back and forth.
GNUstep getting ObjC++ is a complete contribution from Apple, along with a complete rewrite of bison for ObjC/ObjC++.
Links? Or are you just making general assumptions based on one project (KHTML) which they fixed two years ago.
to just give a good example of their innovation which has been given back, and which is considered to be a serious advance: launchd
And no, I'm not going to get into a programming debate about the exact meaning or value of things such as this, because, quite frankly, I have no interest in doing so. I haven programmed in about ten years, so while I do understand a fair amount of what is going on today, I'm too rustry to argue the esoterics anymore.
But here's Apple's developer page on it:
http://developer.apple.com/macosx/launchd.html
Here is the Google page for links, if you are interested, and if the page carries through:
http://www.google.com/search?client=...UTF-8&oe=UTF-8
Some of it is rewriting, and some of it is adding to.
Apple's zerconf, Bonjour is another contribution. Avahi is an example GPL project leveraging the technical ideas of Bonjour.
http://en.wikipedia.org/wiki/Bonjour_(software)
http://en.wikipedia.org/wiki/Avahi_(software)
Lets be really clear, I know the guy who wrote a large chunk of freebsd. The only company contributing to it much any more is Yahoo, who still has a large effort to contribute back.
The truth with freebsd/openbsd/netbsd is that Apple's code is largely unchanged from the kernel, people always say it's "rewritten by apple", but really the didn't do much to the kernel.
Pretty much left FreeBSD's kernel alone and replaced most of it with Mach is what they did. Split it at the upper/lower demarkation and only kept the FreeBSD upper half. That's why Apple changes to the kernel mostly don't make it back to FreeBSD, they are meaningless without the lower Mach portion. Especially since the lock granularity changes have been implemented.
I think Apple has been playing nice with the user-land stuff since it's kernel agnostic.
Wow, glass screens on a phone. That's a brilliant "reliable source" you've got. Did he also tell you that the next MBPro would be made of balsa wood to lower the weight?
So, do you believe me now?
http://www.apple.com/pr/library/2007/06/18iphone.html