I don't know. What do you guys think about this. I'm sure it might be legitimate in some ways, but what if Newton had patented calculus? After all, differentiation is esssentially an algorithm, no? Is it legitimate? Not legal, but ethical?
A patent gives you the right to exploit your technological innovation for a limited time. So a valid use of the MP3 patent would be for Frauenhofer to have licensed the MP3 technology from the beginning so that people making decoders would pay them the royalty. In theory, the product is so good, that people will pay the extra $0.75 for their decoders. Now, if the product isn't good enough to get people to pay for it, then it is unethical to give it away free just so it gets widely adopted and THEN start enforcing the patent. If people had known the technology wouldn't have been free to begin with, they could have used something else. It is very well legal, but its still unethical. Now, remember, patents exist for the good of the consumer, so they get new and better technologies. Only indirectly are the there so companies can get money, the thinking being that if companies make money, then they'll bring new technologies to the people. If companies are just making money, rather than bringing new and good technologies to the market, then the purpose of the patent system is not being served. This is what Frauenhofer is doing.
People don't understand the software. Thus, they pay more in hopes that they'll get a better product. And this isn't just limited to neophytes. Most people do it all the time. For example, say you go to buy a cellphone. Most people don't read reviews of the different cell-phone models before they buy. So what do they do? Do they buy the cheapest model there? Of course not. They'll usually pony up an extra $30-$40 bucks in hopes the get a "good" one.
I was running with 256MB of RAM, with every tweek commonly known (and some not so commonly known). The problem is that you have a different definition of usable than I do. "Truely usable" for me means instant. Things should pop up the nanosecond I click my mouse. There should be *no* visible delay. There should never be any flicker, even when resizing complex windows. BeOS was like that for me on my old 300MHz P300. Take, for example, windows with large bitmaps in them (like the initial screen in the KDE control panel). There is visible rubber-banding when you resize that panel quickly, even on my 2GHz machine. In comparison, I never saw rubber-banding in any native BeOS apps, and that was with only 300Mhz. Or take the example of starting up complex dialog boxes like the Konqueror preferences menu. In BeOS, you clicked the menu item and it appeared. In Konqueror, there is a noticible delay (maybe 1/2 a second) before the menu appears. In all, I lose maybe a whole 30 seconds over the course of a day to UI lag. Not a big deal overall, but it bugs the hell out of me while I'm using the machine.
PS> And yes, I notice flicker (faintly) even at 85Hz and encode my MP3s at 256kbps and they still don't sound right. Funny, though, I can play games at 20fps (Quake III on a P2-300) and not notice the difference...
That's my point. I bought my laptop with only 128MB because Dell charges a ridiculous amount for RAM. After I upgraded the machine to 640MB, it ran fine. But my point is that if a 2GHz machine with 128MB of RAM won't run XP well, a K6 with 128MB of RAM certainly won't run XP well. Yet, some people still think XP on that config runs fine.
Its not so much the 3.1 compiler is buggy, its just that there was a problem in the ABI so they had to change it. Stuff compiled with 3.1 (like KDE 3.0, for example) ran fine on my Gentoo box until I upgraded to 3.2.
BTW> 3.2 is a worthwhile upgrade for all those teetering between upgrading or not to the new GCC 3.2 based distros (RedHat 8.0, Mandrake 9.0). Especially with custom compiled software (optimized for P4 or Athlon) it makes a visible performance difference in the GUI.
Apps don't work together - it's aweful. >>>>>> I hear vague stuff like this all the time. I think its just Mac users who can't get used to something different. I know, personally, using WinXP on the occasions I have to is painful, since I simply don't like the Windows ways of doing things. Yet, I wonder how much of it is real and how much is perceived. Exactly *what* don't you like about Linux (probably with KDE 3.0, given that its the most advanced Linux DE out right now). Like as in list form, referring to specific applications.
.... Um. I may have missed something, but isn't this the way all software works? >>>>>> Nope. Windows XP doesn't work like that. I had everything setup for my network at one point. I don't remember the details, but I ran some wizard to change my computer's identification name, and it changed my IP and DNS addresses behind my back! Took me an hour to figure out why my networking wasn't working anymore.
OS X, 10.1 runs fine, if a bit sluggish on my 9500. To hear people complaining about its performance on G4s makes me laugh. I don't buy it-- I think this is just an excuse from people who are too grumpy to switch from OS 9. >>>>>>>> You're just one of those people who'se reflexes are so slow, that you're technically dead. I've heard some people say that WinXP ran just fine on their K6 400Mhz with 128MB of RAM! WinXP doesn't run okay on my 2GHz laptop with 128MB of RAM (swapping). Hell, KDE 3.0 only became truely usable for me when I upgraded from my 1.5GHz Athlon to my current 2 GHz P4.
A gentoo install does take in the neighborhood of two days. It takes about 3-4 hours to bootstrap and merge the base system (something most people leave on overnight) and upwards of 6-7 hours (because of all the C++) to merge the full KDE desktop. If you add in some actual applications here (like KOffice, another hour or so) or OpenOffice (something like 18 hours) then yeah, you can easily blow two days on it.
BTW> That FVWM comment is crap. FVWM doesn't have 1/4 of 1/8 of 1/16 the power of KDE 3.x. Besides the eye candy (which is nice mind you) FVWM (and the traditional X applications one tends to use with it) have nowhere near the level of integration and polish as KDE desktop apps. Besides, they lack anti-aliasing, which (on my 1600x1200 LCD screen) makes them look horrid beyond comparison. It is true that GNOME and KDE are pretty bloated, but compare them instead to something like Windows 2000, which is comparable in features but lightning fast.
You still haven't given me a good defense of Mac OS X's design. Do you have one or not. Can you refute any of my points? I'll make it easy for you:
1) The single-server design mixes the problems of microkernels (message passing overhead) with the problems of macrokernels (all code in one address space).
2) Mach has a history of slow message passing, which exacerbates #1
3) Quartz Extreme only accelerates eye candy (window transparency and transition effects) instead of real drawing. Given that it uses OpenGL, which supports beziers just fine, it's severely limited. And you know why? Because Apple tied its system too closely to Display PDF.
If you can't refute or answer to any of these points, don't bother replying.
Unfortunately, there are no lmbench numbers for OS X, and I don't have one handy. But most reviews indicate that OS X is quite slow. That, in conjunction with Mach's history of being rather slow, gives me good reason to make the statement that OS X is slow by design.
Hmm, I'm something of an beginning OS buff. I've studied the kernels of Solaris, Linux, Windows, and various BSDs. Kernel devel books litter my bedroom (Modern Operating Systems, Solaris Internals, Design & Implementation of BSD 4.4, Understanding the Linux Kernel, among others). I've designed my own page allocator, written my own copy of the slab allocator, have reverse engineered the method Intel C++ 5.0 uses to call static global constructors and implement exception handling, all in preperation for writing my own kernel. I've got a design spec for the VM down, and I'm in the middle of writing a filesystem. So, no, I'm not a 'leet kernel hacker, but I know something about OS design, and feel experienced enough to criticize Apple's design decisions. You have to remember that OS X is based on NeXTStep, which was designed at the height of the microkernel/component/OS personality craze, and Mach has been widely criticized as a slow microkernel. Nobody at Apple has given a technical rationale for OS X's design, the only thing we ever see is marketing bull catered at mainsteam (dumb) users. I'm not saying that OS X is necessarily a bad OS. I'm just pointing out that its not well designed like the original poster said.
It works. >>>>>> So do modern versions of Windows (2000 and XP) but that doesn't make them well-designed operating systems. Lots of things work that aren't well designed. OS X is quite slow and unstable as far as UNIXs go. A lot of that has to do with bad system design.
You care to give a legitimate defense of OS X's system structure? I don't much like microkernels myself, but Be's design was a hell of a lot cleaner than Apple's amorphous "design." Reminds me more of Windows than a real OS. And note I said OS (not userland) so refrain from making any comments about nifty icons and ease of use.
But I guess you don't even know the difference between "objprelink" and "prelink". >>>>>>> A bit presumptous of you don't you think? I'll bet I've hacked around with prelink and object than you have. This includes running it through debug sessions under gdb and getting it working on my redhat rawhide install. Be a little more tactful in the future.
As for objprelink, you should know that it has been broken for most of the 3.0 series. I was specifically referring to this post on dot.kde.org:
Re: 3.0.3 upgrade? by fredan on Tuesday 20/Aug/2002, @12:29
Nobody is saying anything about the new gcc release.
I can say something. Yesterday I compiled 3.0.3 with gcc 3.1.1 with options like -march=athlon -mmmx -md3now and they have also enabled objprelink again, so it's fast!
And I'd remind you that, outside the technical definition of hypocracy, it is entirely legitimate to accuse a group of hypocracy. Groups have certain ideas, goals, and viewpoints. Belonging to a group associates an individual with those ideas, goals, and viewpoints. Unless the individual points out otherwise, it is often valid to attribute the afformentioned ideas, goals to him. Sometimes it can get you in trouble, such as attributing unsupportable claims on racial groups, but often it works. For example, if certain European companies where illegally dumping waste in rivers, at the same time when the EU was making a big deal about environmental protection, that would be hypocritical of the given European country. Note, that a country is neither an individual nor a group with a defined mission, but it is an entity with cleary defined bounderies, and can legitimately be accused of hypocracy.
I was tempted to do a sarcastic post about Sigma being a genuine American company bringing real dollars into the US economy, but decided that it had been done already:) Still, I'd like to vent about what I perceive as a giant hipocracy on the part of corporate America. They define "stealing" to include making backup copies, but feel perfectly free to actually infringe on other people's copyrights.
Umm, MacOS X is kludgy melange of Mach, FreeBSD, and a candy Apple coating. The single server BSD design is utterly brain-dead and gets rid of the main advantages of a microkernel (seperating kernel code into protected tasks). It is in no way a good operating system. Now the GUI on top is another matter. I'm pissed that Apple didn't go the whole way with the whole vector GUI (they store Windows as bitmaps instead of vector representations as they should be) and think that QE sucks because it just accelerates compositing and some window-level effects instead of actual Quartz drawing, but it can be argued that it is one of the more user-friendly UIs available.
How is it the fault of closed source drivers that he can't ugprade to an AGP card? Even if the drivers were open, he still couldn't upgrade to a TNT2. Do you see my point? He blames closed source drivers for being a limited time use thing. Presumably, NVIDIA might abandon support for certain products (although, as of yet, they still support even the TNT-1, which is the oldest card using the same programming interface) which might necessitate him upgrading. My point was that he still had the older drivers, so he didn't need to upgrade unless he wanted newer drivers. Given that it is rare that devices that old get any updates in newer driver releases, its pointless for him to want newer drivers. In fact, newer drivers tend to be worse for extremely old cards like the TNT. Now, the only valid point he makes is that, on Linux, closed drivers tie him to particular kernels. However, the kernel wrapper is Open Source, so this problem is mitigated.
Umm, your comments make no sense. There is no AGP slot on your computer, so it can't accept an NVIDIA card. Open Source drivers would help you how? If you're talking about NVIDIA dumping support for older cards in newer releases of the binary drivers which (presumably) support newer kernel versions or whatnot
A) A RivaTNT2 (minimum reqs to get into NVIDIA driver land) is $20. How poor are you?
B) The kernel wrapper is Open Source. You can modify it to run on whatever kernel you want. A bunch of people hacked it to make it run on FreeBSD, and NVIDIA techs even provided support!
Slashdot Mirror
I don't know. What do you guys think about this. I'm sure it might be legitimate in some ways, but what if Newton had patented calculus? After all, differentiation is esssentially an algorithm, no? Is it legitimate? Not legal, but ethical?
A patent gives you the right to exploit your technological innovation for a limited time. So a valid use of the MP3 patent would be for Frauenhofer to have licensed the MP3 technology from the beginning so that people making decoders would pay them the royalty. In theory, the product is so good, that people will pay the extra $0.75 for their decoders. Now, if the product isn't good enough to get people to pay for it, then it is unethical to give it away free just so it gets widely adopted and THEN start enforcing the patent. If people had known the technology wouldn't have been free to begin with, they could have used something else. It is very well legal, but its still unethical. Now, remember, patents exist for the good of the consumer, so they get new and better technologies. Only indirectly are the there so companies can get money, the thinking being that if companies make money, then they'll bring new technologies to the people. If companies are just making money, rather than bringing new and good technologies to the market, then the purpose of the patent system is not being served. This is what Frauenhofer is doing.
People don't understand the software. Thus, they pay more in hopes that they'll get a better product. And this isn't just limited to neophytes. Most people do it all the time. For example, say you go to buy a cellphone. Most people don't read reviews of the different cell-phone models before they buy. So what do they do? Do they buy the cheapest model there? Of course not. They'll usually pony up an extra $30-$40 bucks in hopes the get a "good" one.
I was running with 256MB of RAM, with every tweek commonly known (and some not so commonly known). The problem is that you have a different definition of usable than I do. "Truely usable" for me means instant. Things should pop up the nanosecond I click my mouse. There should be *no* visible delay. There should never be any flicker, even when resizing complex windows. BeOS was like that for me on my old 300MHz P300. Take, for example, windows with large bitmaps in them (like the initial screen in the KDE control panel). There is visible rubber-banding when you resize that panel quickly, even on my 2GHz machine. In comparison, I never saw rubber-banding in any native BeOS apps, and that was with only 300Mhz. Or take the example of starting up complex dialog boxes like the Konqueror preferences menu. In BeOS, you clicked the menu item and it appeared. In Konqueror, there is a noticible delay (maybe 1/2 a second) before the menu appears. In all, I lose maybe a whole 30 seconds over the course of a day to UI lag. Not a big deal overall, but it bugs the hell out of me while I'm using the machine.
PS> And yes, I notice flicker (faintly) even at 85Hz and encode my MP3s at 256kbps and they still don't sound right. Funny, though, I can play games at 20fps (Quake III on a P2-300) and not notice the difference...
That's my point. I bought my laptop with only 128MB because Dell charges a ridiculous amount for RAM. After I upgraded the machine to 640MB, it ran fine. But my point is that if a 2GHz machine with 128MB of RAM won't run XP well, a K6 with 128MB of RAM certainly won't run XP well. Yet, some people still think XP on that config runs fine.
Its not so much the 3.1 compiler is buggy, its just that there was a problem in the ABI so they had to change it. Stuff compiled with 3.1 (like KDE 3.0, for example) ran fine on my Gentoo box until I upgraded to 3.2.
BTW> 3.2 is a worthwhile upgrade for all those teetering between upgrading or not to the new GCC 3.2 based distros (RedHat 8.0, Mandrake 9.0). Especially with custom compiled software (optimized for P4 or Athlon) it makes a visible performance difference in the GUI.
Apps don't work together - it's aweful.
>>>>>>
I hear vague stuff like this all the time. I think its just Mac users who can't get used to something different. I know, personally, using WinXP on the occasions I have to is painful, since I simply don't like the Windows ways of doing things. Yet, I wonder how much of it is real and how much is perceived. Exactly *what* don't you like about Linux (probably with KDE 3.0, given that its the most advanced Linux DE out right now). Like as in list form, referring to specific applications.
.... Um. I may have missed something, but isn't this the way all software works?
>>>>>>
Nope. Windows XP doesn't work like that. I had everything setup for my network at one point. I don't remember the details, but I ran some wizard to change my computer's identification name, and it changed my IP and DNS addresses behind my back! Took me an hour to figure out why my networking wasn't working anymore.
OS X, 10.1 runs fine, if a bit sluggish on my 9500. To hear people complaining about its performance on G4s makes me laugh. I don't buy it-- I think this is just an excuse from people who are too grumpy to switch from OS 9.
>>>>>>>>
You're just one of those people who'se reflexes are so slow, that you're technically dead. I've heard some people say that WinXP ran just fine on their K6 400Mhz with 128MB of RAM! WinXP doesn't run okay on my 2GHz laptop with 128MB of RAM (swapping). Hell, KDE 3.0 only became truely usable for me when I upgraded from my 1.5GHz Athlon to my current 2 GHz P4.
A gentoo install does take in the neighborhood of two days. It takes about 3-4 hours to bootstrap and merge the base system (something most people leave on overnight) and upwards of 6-7 hours (because of all the C++) to merge the full KDE desktop. If you add in some actual applications here (like KOffice, another hour or so) or OpenOffice (something like 18 hours) then yeah, you can easily blow two days on it.
BTW> That FVWM comment is crap. FVWM doesn't have 1/4 of 1/8 of 1/16 the power of KDE 3.x. Besides the eye candy (which is nice mind you) FVWM (and the traditional X applications one tends to use with it) have nowhere near the level of integration and polish as KDE desktop apps. Besides, they lack anti-aliasing, which (on my 1600x1200 LCD screen) makes them look horrid beyond comparison. It is true that GNOME and KDE are pretty bloated, but compare them instead to something like Windows 2000, which is comparable in features but lightning fast.
8 spaces? 8-space tabs are for amateurs. Real programmers use 4-tab spaces!
You still haven't given me a good defense of Mac OS X's design. Do you have one or not. Can you refute any of my points? I'll make it easy for you:
1) The single-server design mixes the problems of microkernels (message passing overhead) with the problems of macrokernels (all code in one address space).
2) Mach has a history of slow message passing, which exacerbates #1
3) Quartz Extreme only accelerates eye candy (window transparency and transition effects) instead of real drawing. Given that it uses OpenGL, which supports beziers just fine, it's severely limited. And you know why? Because Apple tied its system too closely to Display PDF.
If you can't refute or answer to any of these points, don't bother replying.
That would be hippocracy. n00b...
Unfortunately, there are no lmbench numbers for OS X, and I don't have one handy. But most reviews indicate that OS X is quite slow. That, in conjunction with Mach's history of being rather slow, gives me good reason to make the statement that OS X is slow by design.
Hmm, I'm something of an beginning OS buff. I've studied the kernels of Solaris, Linux, Windows, and various BSDs. Kernel devel books litter my bedroom (Modern Operating Systems, Solaris Internals, Design & Implementation of BSD 4.4, Understanding the Linux Kernel, among others). I've designed my own page allocator, written my own copy of the slab allocator, have reverse engineered the method Intel C++ 5.0 uses to call static global constructors and implement exception handling, all in preperation for writing my own kernel. I've got a design spec for the VM down, and I'm in the middle of writing a filesystem. So, no, I'm not a 'leet kernel hacker, but I know something about OS design, and feel experienced enough to criticize Apple's design decisions. You have to remember that OS X is based on NeXTStep, which was designed at the height of the microkernel/component/OS personality craze, and Mach has been widely criticized as a slow microkernel. Nobody at Apple has given a technical rationale for OS X's design, the only thing we ever see is marketing bull catered at mainsteam (dumb) users. I'm not saying that OS X is necessarily a bad OS. I'm just pointing out that its not well designed like the original poster said.
It works.
>>>>>>
So do modern versions of Windows (2000 and XP) but that doesn't make them well-designed operating systems. Lots of things work that aren't well designed. OS X is quite slow and unstable as far as UNIXs go. A lot of that has to do with bad system design.
You care to give a legitimate defense of OS X's system structure? I don't much like microkernels myself, but Be's design was a hell of a lot cleaner than Apple's amorphous "design." Reminds me more of Windows than a real OS. And note I said OS (not userland) so refrain from making any comments about nifty icons and ease of use.
Ack, I meant to type objprelink.
But I guess you don't even know the difference between "objprelink" and "prelink".
>>>>>>>
A bit presumptous of you don't you think? I'll bet I've hacked around with prelink and object than you have. This includes running it through debug sessions under gdb and getting it working on my redhat rawhide install. Be a little more tactful in the future.
As for objprelink, you should know that it has been broken for most of the 3.0 series. I was specifically referring to this post on dot.kde.org:
Re: 3.0.3 upgrade?
by fredan on Tuesday 20/Aug/2002, @12:29
Nobody is saying anything about the new gcc release.
I can say something. Yesterday I compiled 3.0.3 with gcc 3.1.1 with options like -march=athlon -mmmx -md3now and they have also enabled objprelink again, so it's fast!
And I'd remind you that, outside the technical definition of hypocracy, it is entirely legitimate to accuse a group of hypocracy. Groups have certain ideas, goals, and viewpoints. Belonging to a group associates an individual with those ideas, goals, and viewpoints. Unless the individual points out otherwise, it is often valid to attribute the afformentioned ideas, goals to him. Sometimes it can get you in trouble, such as attributing unsupportable claims on racial groups, but often it works. For example, if certain European companies where illegally dumping waste in rivers, at the same time when the EU was making a big deal about environmental protection, that would be hypocritical of the given European country. Note, that a country is neither an individual nor a group with a defined mission, but it is an entity with cleary defined bounderies, and can legitimately be accused of hypocracy.
I heared the re-enabled prelink in 3.0.3. In idea why, and does it also work in 3.1 beta 1?
I was tempted to do a sarcastic post about Sigma being a genuine American company bringing real dollars into the US economy, but decided that it had been done already :) Still, I'd like to vent about what I perceive as a giant hipocracy on the part of corporate America. They define "stealing" to include making backup copies, but feel perfectly free to actually infringe on other people's copyrights.
Umm, MacOS X is kludgy melange of Mach, FreeBSD, and a candy Apple coating. The single server BSD design is utterly brain-dead and gets rid of the main advantages of a microkernel (seperating kernel code into protected tasks). It is in no way a good operating system. Now the GUI on top is another matter. I'm pissed that Apple didn't go the whole way with the whole vector GUI (they store Windows as bitmaps instead of vector representations as they should be) and think that QE sucks because it just accelerates compositing and some window-level effects instead of actual Quartz drawing, but it can be argued that it is one of the more user-friendly UIs available.
When was it that you last checked? The last update was two weeks ago.
How is it the fault of closed source drivers that he can't ugprade to an AGP card? Even if the drivers were open, he still couldn't upgrade to a TNT2. Do you see my point? He blames closed source drivers for being a limited time use thing. Presumably, NVIDIA might abandon support for certain products (although, as of yet, they still support even the TNT-1, which is the oldest card using the same programming interface) which might necessitate him upgrading. My point was that he still had the older drivers, so he didn't need to upgrade unless he wanted newer drivers. Given that it is rare that devices that old get any updates in newer driver releases, its pointless for him to want newer drivers. In fact, newer drivers tend to be worse for extremely old cards like the TNT. Now, the only valid point he makes is that, on Linux, closed drivers tie him to particular kernels. However, the kernel wrapper is Open Source, so this problem is mitigated.
That's more that the x86 has certain demands of the stack rather than C.
Umm, your comments make no sense. There is no AGP slot on your computer, so it can't accept an NVIDIA card. Open Source drivers would help you how? If you're talking about NVIDIA dumping support for older cards in newer releases of the binary drivers which (presumably) support newer kernel versions or whatnot
A) A RivaTNT2 (minimum reqs to get into NVIDIA driver land) is $20. How poor are you?
B) The kernel wrapper is Open Source. You can modify it to run on whatever kernel you want. A bunch of people hacked it to make it run on FreeBSD, and NVIDIA techs even provided support!