So the obvious answer would be to move one of the processes to the other core.
The obvious answer is that the situation isn't a whole lot different from dual processor machines...
Re:"Paltry" is probably a poor choice of words
on
GCC 4.0.0 Released
·
· Score: 1
I don't wish to start a flamewar, zealots of both sides please go play elsewhere, but Apple probably has more users than Linux, and every other Unix box combined is probably dwarfed by Linux.
I don't believe that's true. Even looking at just desktops (but including business and academic desktops), Linux seems to be doing better than OS X. The only areas where OS X does better than Linux seems to be home computers and laptops (with some desktop Linux users using Macintosh laptops).
Around here, Macintosh has essentially disappeared completely with not even any stores with 60 miles carrying Apple or Macintosh software anymore, but Linux is growing strongly.
But if you have numbers that say otherwise, maybe you can share them?
Zenph Studios, a software company based in Raleigh, North Carolina, claims that it has found a solution to the problem, although it refuses to say how for commercial reasons.
There have been attempts at music transcription software since the 1970's. For some obscure reason, many of the people who tried didn't seem to think of the fact that classical music comes with a score.
So, the "solution" to the problem is simple: use the known score to get the notes and polyphony, and use the recording primarily to infer the performance parameters. It's not a very complicated problem, actually. I suspect the main reason why you haven't seen this before is because it is of fairly little commercial interest.
If you really don't understand the difference between transitioning to kernel mode and transitioning/sharing data with a user mode server, I don't think you have sufficient understanding of the subject to discuss it in a meaningful way with those of us who do. Sorry.
Of course, there are "differences". Of course, putting stuff into the kernel makes it easier for the NT graphics server to access user mode data structures. But if you don't understand why that isn't relevant in this context (e.g., you still need conversions between hardware independent bitmaps into hardware dependent ones), then you really don't have a sufficient understanding to discuss the matter in a meaningful way with people who do.
The rest of your post is generally wrong and/or irrelevant to the issues at hand.
Well, you don't know what you're talking about. But don't feel bad about it--you share that with a lot of people in this industry. After all, all the flaky, bloated, shitty software that ships must come from somewhere, and it comes from people with misconceptions like yours.
Just keep in mind is that Microsoft is now on at least the fourth design of graphics subsystem (95/98, NT-user-mode, NT-kernel-mode, now Avalon). Apple had to throw out their first window system and buy themselves a new one. Sun tried twice (SunView, NeWS) with a new window system and failed both times. BeOS didn't succeed either. Neither did the AT&T Blit, or DPS, or any of the others.
The X window system design, on the other hand, has stood the test of time, with huge longevity, numerous implementations, excellent scalability, and excellent performance. Evidently, the people who designed X got it right more than 20 years ago. The fact that you don't understand what they achieved or how they did it is only your loss.
For version 4.0, released in 1996, the graphics components were moved from a user-mode module to a kernel-mode module
True, but the point is: it is running in a different context from the application.
Moving the graphics components from a server process to a kernel module was a good trade-off on Windows
No, it wasn't a "good tradeoff", it was a way of partially mitigating the consequences of trying to continue to hold on to an obsolete and poorly designed graphics API in Windows. Microsoft needed X like client/server separation, but they either didn't even realize what they were doing, or they weren't willing to do the engineering necessary to design a decent protocol and create a new API to do it. And Microsoft has paid a heavy price for it in increased kernel complexity and reduced kernel stability.
Moving X into a kernel-mode module on Linux would be a much less favourable trade-off because,
Moving X into a kernel module wouldn't help you because X was designed for client/server operations from the ground up. You don't gain any significant amount of efficiency by moving the X server into the kernel. In addition, the Linux kernel appears to handle communications between user processes more efficiently than the NT kernel as well, further reducing the need to put anything into the kernel relative to NT. That's why the question has never come up.
Yes, the US can enforce laws out further than that. On the other hand, you don't need a work visa if you work on a ship that just happens to anchor in territorial waters.
How this falls out depends on what politicians make of it. They can prohibit this sort of conduct and send the coast guard to send the ship packing, or they can actually view it as a reasonable political compromise that doesn't force them to touch the H1b issue one way or another.
I suspect that, if this ever were to become a big thing commercially, inaction and silent toleration would be the preferred course for most politicians. Only if it looked like it became a media debacle would they likely start acting.
they're basically prisoners on the ship unless they can manage to get a Visa to enter the country. Which, of course, negates the entire point of not messing with H1-Bs.
These people can get visitor visas just fine. They can go to shore pretty much whenever they like and go shopping, have girl/boyfriends, and do all those things Americans like to do in their free time. The only thing they cannot do in the US is work there and pay taxes there.
The fact that they can work 3 miles outside the US and visit whenever they like only demonstrates the absurdity of placing caps on US work visas.
With insightful analysis like yours, I'm not surprised companies have to outsource...
And what VISA are they going to use to gain enterance to the US?
A visitor visa. There are no caps on visitor visas and they are easy to get. The restriction is that you can't work on them in the US, but these people won't be working in the US--that's the point.
The "11" is the major version of the protocol; it changes if the protocol changes incompatibly. If the protocol doesn't change incompatibly, then the major version doesn't change. Right now, it doesn't look like the protocol will ever need to change incompatibly anymore, so X11 it is.
Sorry, no go. You can make graphics appear remotely, but that isn't network transparent windowing. The reason Jobs nixed the remote parts of NeXT in OS X is probably because he realized that.
So in another 4-5 or more years X will have the same stuff that OS X has had for a while?
You are confusing visual fluff with functionality. The fact is that X already has more than OS X will likely ever have.
This highlights the problem with Opensourcesoftwaredevelopmenten.
It took Apple until the 21st century to offer reliable multitasking, a reasonable kernel, and a decent file system, and they managed to do that only by (1) making extensive use of open source software and (2) buying the stuff outside. Apparently, Apple has trouble doing the "un-fun stuff" if it takes them this long. See, that's the problem with closed source.
Re:How about doing something actually useful ?
on
Next Generation X11
·
· Score: 1
How about implementing dynamic X server reconfiguration to allow connecting and disconnecting external monitors to laptops on the fly? How about using different resolutions on these monitors?
That's been there for, oh, at least a decade in XFree86. Of course, other X11 servers have had it longer.
Right now Linux/X11 is horribly behind both Windows and Mac OS X, being unable to detect an external monitor being connected and change resolution accordingly.
but there's no two ways about it: the X server is too old and archaic for modern interfaces.
How do you think Windows or Macintosh work? All major GUIs use an X11-like architecture these days, with separate server and client processes talking via IPC.
I could be talking out of my ass here, but I would expect that what one would REALLY want is abstration at the DISPATCHING level... use some mechanism to dispatch the communication between the client and server, then an OS with no need for networking could use X by replacing the implementation with
X11 doesn't require networking; it runs fine over non-network IPC mechanisms and is commonly run that way.
These aren't necessarily opposing opinions: the graphics system is different than a kernel. But in a GUI environment, both are going to be responsible for a *lot* of communication, so I find it intriguing that people are able to hold both opinions at once.
What can I say? The design of the Linux kernel sucks. Tanenbaum was right. But the Linux kernel works, so why worry about it? The fact that it isn't a microkernel is largely Linus's headache, and as long as he can live with it, nobody is going to worry about it.
And what is happening with X11 and Linux is the same that's happening on Windows and Macintosh: all of those have a graphics server process and a monolithic kernel. Microsoft and Apple came to those designs after trying a lot of other possibilities, so there is probably a reason for it.
Re:Why isn't this already out?
on
Next Generation X11
·
· Score: 2, Informative
Well actually, yes it does. You still have to marshal pixmaps (a wretched and primitive yet still bloated format) into a shared memory segment just so the server can pull them out of there and transfer them to the graphics driver.
How the hell do you think this works on, say, Macintosh or Windows? Magic elves make the bits appear on your screen? X11 gives you the same choice as any other system: you can either let the server do a copy-conversion, or you can use direct rendering. Most sane people will choose letting the server do the copy for most applications.
And X's implementation of network transparency doesn't give clients any way to tell the server to aggregate events or even tell it to "shut the hell up already" with all the mouseover events over regions where they're not listening to the events.
Huh? You have detailed control over what regions you get what events for. But what's the point anyway these days? Mouse events are not a performance bottleneck on X11.
Network transparency is a good thing. X's implementation of it stinks.
First of all, X11 isn't an "implementation", it's a protocol
Secondly, implementations of X11, like XFree86, do pretty much the same thing Macintosh or Windows do: they have a graphics server process that's separate from the application, and they use IPC to talk to it.
The only thing that "stinks" is that people like you who evidently have no idea what's going on opine about X11's design.
I think the real problem is, everyone knows X Windows is broken, but nobody knows what to do about it.
There is nothing to be done. Despite the persistent put-downs from Apple and Windows fan-boys, despite attacks from every proprietary vendor of window systems under the sun since day one, the X Window System (which is actually a protocol) has outlived NeWS, SunView, NeXT, MacOS, and just about every other window system there ever was. It is supported by many vendors and has multiple independent implementations. Chances are, X11 will be around long after OS X, Aqua, and Quartz will be dim memories.
As a Mac OS X user, a previous Windows user, and a current Linux Desktop user, I will not be the first to tell you that X is slooow.
On comparable hardware, X11 runs rings around both Windows and OS X.
Lastly, the problem comes with there being absolutely no good drivers available. Honestly, even though NVidia/ATi tries, they're not up to par with what they've got on the Windows platform, and Apple developers have had the luxury of seeing the developer's specs, so their drivers are just as impecible.
X11 isn't just XFree86. X11 has had commercial, hardware accelerated drivers long before OS X even came out.
You can use shared memory as a transport, just like any other. It turns out not to be any faster than UNIX domain sockets.
People have also hacked Xlib to basically interface to a drawing library. But that just turns out to be not very useful (and it isn't really fast either).
I don't know much about the guts of X and windowing in general, but what exactly is the benefit of having windowing in an external process?
It makes things work better. Microsoft thought they could get away without it and regretted it. Apple thought they could get away without it and regretted it.
Are interoperability and backwards compatibility the principal reasons for X?
No, the principal reasons are that the design works really well and has stood the test of time. The designs you propose were the designs people started out with 30 years ago.
Personally, I don't care about the network transparency of X - I'd prefer the remote capabilities to be built on top of the existing drawing system as remote use seems to me
With municipal WiFi, we, the people, actually get to control it. With private Internet access, companies can make arbitrary rules. They happen not to restrict content too much right now, but that's not guaranteed to last. And if it's companies doing it, you don't have any recourse.
As a parent, you cannot, indeed should not, be by your teenagers side 24/7.
Quite right. So, you should just have brought them up such that when they are 15, they either heed your rule that they don't chat at all, or they are mature enough to handle it.
If your 15 year old chats on AOL and makes sex dates with 28 year olds, then that's a problem with how you raised her, not AOL. AOL monitors aren't going to help you there.
To get OS X-like redrawing (namely, completely smooth moves) use the "-wm" option to the X server. This will keep an in-memory copy of whatever window contents are present. It should eliminated damaged window contents, etc. This may be a good idea for the permanently complaining OS X crowd and those who whine about X11 redrawing.
You can tune other system paging behavior with "sysctl" (maybe increasing the swappiness is what you want). Note that Linux is already faster than Panther in those systems-y areas, despite whatever Panther does with caching, so I'm not sure what you would hope to achieve with this. (See here.)
A smooth uncluttered interface that works as expected with a minimum of tweaking, and features like graphics accelerated Exposé rank a bit higher than just theming, don't they?
I challenge you to show evidence that the OS X interface is significantly "better" in any quantifiable sense than Windows, Gnome, or KDE.
It's like saying a Mercedes is just about having fancier trim.
So, yes, your analogy seems apt: Mercedes, like Macintosh, is a premium-priced status symbol. If you want value, quality, functionality, or innovation, there are better choices.
Slashdot Mirror
So the obvious answer would be to move one of the processes to the other core.
The obvious answer is that the situation isn't a whole lot different from dual processor machines...
I don't wish to start a flamewar, zealots of both sides please go play elsewhere, but Apple probably has more users than Linux, and every other Unix box combined is probably dwarfed by Linux.
I don't believe that's true. Even looking at just desktops (but including business and academic desktops), Linux seems to be doing better than OS X. The only areas where OS X does better than Linux seems to be home computers and laptops (with some desktop Linux users using Macintosh laptops).
Around here, Macintosh has essentially disappeared completely with not even any stores with 60 miles carrying Apple or Macintosh software anymore, but Linux is growing strongly.
But if you have numbers that say otherwise, maybe you can share them?
Zenph Studios, a software company based in Raleigh, North Carolina, claims that it has found a solution to the problem, although it refuses to say how for commercial reasons.
There have been attempts at music transcription software since the 1970's. For some obscure reason, many of the people who tried didn't seem to think of the fact that classical music comes with a score.
So, the "solution" to the problem is simple: use the known score to get the notes and polyphony, and use the recording primarily to infer the performance parameters. It's not a very complicated problem, actually. I suspect the main reason why you haven't seen this before is because it is of fairly little commercial interest.
If you really don't understand the difference between transitioning to kernel mode and transitioning/sharing data with a user mode server, I don't think you have sufficient understanding of the subject to discuss it in a meaningful way with those of us who do. Sorry.
Of course, there are "differences". Of course, putting stuff into the kernel makes it easier for the NT graphics server to access user mode data structures. But if you don't understand why that isn't relevant in this context (e.g., you still need conversions between hardware independent bitmaps into hardware dependent ones), then you really don't have a sufficient understanding to discuss the matter in a meaningful way with people who do.
The rest of your post is generally wrong and/or irrelevant to the issues at hand.
Well, you don't know what you're talking about. But don't feel bad about it--you share that with a lot of people in this industry. After all, all the flaky, bloated, shitty software that ships must come from somewhere, and it comes from people with misconceptions like yours.
Just keep in mind is that Microsoft is now on at least the fourth design of graphics subsystem (95/98, NT-user-mode, NT-kernel-mode, now Avalon). Apple had to throw out their first window system and buy themselves a new one. Sun tried twice (SunView, NeWS) with a new window system and failed both times. BeOS didn't succeed either. Neither did the AT&T Blit, or DPS, or any of the others.
The X window system design, on the other hand, has stood the test of time, with huge longevity, numerous implementations, excellent scalability, and excellent performance. Evidently, the people who designed X got it right more than 20 years ago. The fact that you don't understand what they achieved or how they did it is only your loss.
For version 4.0, released in 1996, the graphics components were moved from a user-mode module to a kernel-mode module
True, but the point is: it is running in a different context from the application.
Moving the graphics components from a server process to a kernel module was a good trade-off on Windows
No, it wasn't a "good tradeoff", it was a way of partially mitigating the consequences of trying to continue to hold on to an obsolete and poorly designed graphics API in Windows. Microsoft needed X like client/server separation, but they either didn't even realize what they were doing, or they weren't willing to do the engineering necessary to design a decent protocol and create a new API to do it. And Microsoft has paid a heavy price for it in increased kernel complexity and reduced kernel stability.
Moving X into a kernel-mode module on Linux would be a much less favourable trade-off because,
Moving X into a kernel module wouldn't help you because X was designed for client/server operations from the ground up. You don't gain any significant amount of efficiency by moving the X server into the kernel. In addition, the Linux kernel appears to handle communications between user processes more efficiently than the NT kernel as well, further reducing the need to put anything into the kernel relative to NT. That's why the question has never come up.
Yes, the US can enforce laws out further than that. On the other hand, you don't need a work visa if you work on a ship that just happens to anchor in territorial waters.
How this falls out depends on what politicians make of it. They can prohibit this sort of conduct and send the coast guard to send the ship packing, or they can actually view it as a reasonable political compromise that doesn't force them to touch the H1b issue one way or another.
I suspect that, if this ever were to become a big thing commercially, inaction and silent toleration would be the preferred course for most politicians. Only if it looked like it became a media debacle would they likely start acting.
they're basically prisoners on the ship unless they can manage to get a Visa to enter the country. Which, of course, negates the entire point of not messing with H1-Bs.
These people can get visitor visas just fine. They can go to shore pretty much whenever they like and go shopping, have girl/boyfriends, and do all those things Americans like to do in their free time. The only thing they cannot do in the US is work there and pay taxes there.
The fact that they can work 3 miles outside the US and visit whenever they like only demonstrates the absurdity of placing caps on US work visas.
With insightful analysis like yours, I'm not surprised companies have to outsource...
And what VISA are they going to use to gain enterance to the US?
A visitor visa. There are no caps on visitor visas and they are easy to get. The restriction is that you can't work on them in the US, but these people won't be working in the US--that's the point.
Humanitarily speaking, since they are not actually in any country, who protects the rights of those 600 laboring software engineers?
As long as most jobs aren't handled this way, those people have a choice. If they don't want to do this, they don't have to.
The "11" is the major version of the protocol; it changes if the protocol changes incompatibly. If the protocol doesn't change incompatibly, then the major version doesn't change. Right now, it doesn't look like the protocol will ever need to change incompatibly anymore, so X11 it is.
Sorry, no go. You can make graphics appear remotely, but that isn't network transparent windowing. The reason Jobs nixed the remote parts of NeXT in OS X is probably because he realized that.
So in another 4-5 or more years X will have the same stuff that OS X has had for a while?
You are confusing visual fluff with functionality. The fact is that X already has more than OS X will likely ever have.
This highlights the problem with Opensourcesoftwaredevelopmenten.
It took Apple until the 21st century to offer reliable multitasking, a reasonable kernel, and a decent file system, and they managed to do that only by (1) making extensive use of open source software and (2) buying the stuff outside. Apparently, Apple has trouble doing the "un-fun stuff" if it takes them this long. See, that's the problem with closed source.
How about implementing dynamic X server reconfiguration to allow connecting and disconnecting external monitors to laptops on the fly? How about using different resolutions on these monitors?
That's been there for, oh, at least a decade in XFree86. Of course, other X11 servers have had it longer.
Right now Linux/X11 is horribly behind both Windows and Mac OS X, being unable to detect an external monitor being connected and change resolution accordingly.
No, it isn't.
but there's no two ways about it: the X server is too old and archaic for modern interfaces.
How do you think Windows or Macintosh work? All major GUIs use an X11-like architecture these days, with separate server and client processes talking via IPC.
I could be talking out of my ass here, but I would expect that what one would REALLY want is abstration at the DISPATCHING level... use some mechanism to dispatch the communication between the client and server, then an OS with no need for networking could use X by replacing the implementation with
X11 doesn't require networking; it runs fine over non-network IPC mechanisms and is commonly run that way.
These aren't necessarily opposing opinions: the graphics system is different than a kernel. But in a GUI environment, both are going to be responsible for a *lot* of communication, so I find it intriguing that people are able to hold both opinions at once.
What can I say? The design of the Linux kernel sucks. Tanenbaum was right. But the Linux kernel works, so why worry about it? The fact that it isn't a microkernel is largely Linus's headache, and as long as he can live with it, nobody is going to worry about it.
And what is happening with X11 and Linux is the same that's happening on Windows and Macintosh: all of those have a graphics server process and a monolithic kernel. Microsoft and Apple came to those designs after trying a lot of other possibilities, so there is probably a reason for it.
Well actually, yes it does. You still have to marshal pixmaps (a wretched and primitive yet still bloated format) into a shared memory segment just so the server can pull them out of there and transfer them to the graphics driver.
How the hell do you think this works on, say, Macintosh or Windows? Magic elves make the bits appear on your screen? X11 gives you the same choice as any other system: you can either let the server do a copy-conversion, or you can use direct rendering. Most sane people will choose letting the server do the copy for most applications.
And X's implementation of network transparency doesn't give clients any way to tell the server to aggregate events or even tell it to "shut the hell up already" with all the mouseover events over regions where they're not listening to the events.
Huh? You have detailed control over what regions you get what events for. But what's the point anyway these days? Mouse events are not a performance bottleneck on X11.
Network transparency is a good thing. X's implementation of it stinks.
First of all, X11 isn't an "implementation", it's a protocol
Secondly, implementations of X11, like XFree86, do pretty much the same thing Macintosh or Windows do: they have a graphics server process that's separate from the application, and they use IPC to talk to it.
The only thing that "stinks" is that people like you who evidently have no idea what's going on opine about X11's design.
One might argue that if X were properly designed then the choice of Window manager shouldn't slow down the performance of the windowing system.
And it doesn't. What's your point? Do you even know what a window manager is?
I think the real problem is, everyone knows X Windows is broken, but nobody knows what to do about it.
There is nothing to be done. Despite the persistent put-downs from Apple and Windows fan-boys, despite attacks from every proprietary vendor of window systems under the sun since day one, the X Window System (which is actually a protocol) has outlived NeWS, SunView, NeXT, MacOS, and just about every other window system there ever was. It is supported by many vendors and has multiple independent implementations. Chances are, X11 will be around long after OS X, Aqua, and Quartz will be dim memories.
As a Mac OS X user, a previous Windows user, and a current Linux Desktop user, I will not be the first to tell you that X is slooow.
On comparable hardware, X11 runs rings around both Windows and OS X.
Lastly, the problem comes with there being absolutely no good drivers available. Honestly, even though NVidia/ATi tries, they're not up to par with what they've got on the Windows platform, and Apple developers have had the luxury of seeing the developer's specs, so their drivers are just as impecible.
X11 isn't just XFree86. X11 has had commercial, hardware accelerated drivers long before OS X even came out.
You can use shared memory as a transport, just like any other. It turns out not to be any faster than UNIX domain sockets.
People have also hacked Xlib to basically interface to a drawing library. But that just turns out to be not very useful (and it isn't really fast either).
I don't know much about the guts of X and windowing in general, but what exactly is the benefit of having windowing in an external process?
It makes things work better. Microsoft thought they could get away without it and regretted it. Apple thought they could get away without it and regretted it.
Are interoperability and backwards compatibility the principal reasons for X?
No, the principal reasons are that the design works really well and has stood the test of time. The designs you propose were the designs people started out with 30 years ago.
Personally, I don't care about the network transparency of X - I'd prefer the remote capabilities to be built on top of the existing drawing system as remote use seems to me
You don't pay for it, so don't worry about it.
With municipal WiFi, we, the people, actually get to control it. With private Internet access, companies can make arbitrary rules. They happen not to restrict content too much right now, but that's not guaranteed to last. And if it's companies doing it, you don't have any recourse.
As a parent, you cannot, indeed should not, be by your teenagers side 24/7.
Quite right. So, you should just have brought them up such that when they are 15, they either heed your rule that they don't chat at all, or they are mature enough to handle it.
If your 15 year old chats on AOL and makes sex dates with 28 year olds, then that's a problem with how you raised her, not AOL. AOL monitors aren't going to help you there.
To get OS X-like redrawing (namely, completely smooth moves) use the "-wm" option to the X server. This will keep an in-memory copy of whatever window contents are present. It should eliminated damaged window contents, etc. This may be a good idea for the permanently complaining OS X crowd and those who whine about X11 redrawing.
You can tune other system paging behavior with "sysctl" (maybe increasing the swappiness is what you want). Note that Linux is already faster than Panther in those systems-y areas, despite whatever Panther does with caching, so I'm not sure what you would hope to achieve with this. (See here.)
A smooth uncluttered interface that works as expected with a minimum of tweaking, and features like graphics accelerated Exposé rank a bit higher than just theming, don't they?
I challenge you to show evidence that the OS X interface is significantly "better" in any quantifiable sense than Windows, Gnome, or KDE.
It's like saying a Mercedes is just about having fancier trim.
I dunno, what do you think Mercedes is about? Their quality is below industry average, and they cost a premium for the performance and features they offer.
So, yes, your analogy seems apt: Mercedes, like Macintosh, is a premium-priced status symbol. If you want value, quality, functionality, or innovation, there are better choices.