Yes, although I've had Linux for nearly two years, I'm still fairly new to it. So perhaps I could us e a pointer on LILO. In order to use two Ethernet cards of the same type, I have to specify the IRQ and base address of each on the boot: command line. I tried to put two "ether=" statements in lilo.conf, but it rejects those. I tried to use an "append=" statement with both "ether" parameters in it, but it seems to have no effect, either writing to my hard drive boot sector or to the boot floppy. (I do love having the floppy immediately access the HD based kernel, though.) Any idea what I'm doing wrong?
Regarding DLLs: You are correct, but it's not just apps; most of the OS is implemented in DLLs, too. Get yourself a white-chocolate-macadamia-nut cookie.:-)
So here's my point about modularity: In NT, you don't have to compile hooks into the kernel for every new card that comes down the line. I bought a SCSI card that didn't even exist when my NT kernel was installed, and just plugged the driver right in and rebooted.
In Linux, I had to: 1. Install GCC 2. Install the kernel source (making sure to get the proper version) 3. Download the driver source 4. Patch the kernel sources and makefiles 5. Build the new kernel 6. Install the new kernel 7. Edit lilo.conf and rerun lilo 8. Reboot -- all this just because I used a new kind of SCSI card! Furthermore, I have to repeat this process _EVERY TIME_ I want to upgrade or patch my kernel, since the driver is not part of the standard distro! Bleagh.
That's the essence of free software, isn't it? As a matter of fact, the makers of the SCSI card in question are working to get support built into some distributions (don't ask me why they don't talk to the kernel team). My whole point way back at the beginning of this thread was, SOME operating systems let you add new hardware species without compiling device specific module hooks into the kernel. Like Windows and MacOS. That's all.
Linux didn't give me a choice, either. My SCSI card isn't supported by the standard distribution. That means, lucky me, I have to edit the kernel source code every time I upgrade my system. And that's with a SCSI card that comes with Linux (and NT) drivers; imagine how screwed I'd be if it didn't!
I'm not arguing against the benefits of being able to performance-tune your kernel (although I think it would be better to do that through user-selectable parameters). I'm saying Linux' inability to accept new hardware without explicit kernel support is a crucial flaw. Please refer to my other posts in this thread.
The difference is, I don't have to mess with source code. I don't have to have a C compiler on my system. (What's that save me, 100MB? I can put a lot of stuff in 100MB.) When I feel like adding even a standard, full-package RPM kernel to Linux, I still have to edit kernel source code, just because my SCSI card hasn't blessed by St. Linus yet. Even though the driver supports kernels back to 2.0.27. Even NT doesn't require the kernel to explicitly support each loadable module.
So there's the difference: No source code, no comilation, total separation of drivers from kernel.
Yes, I installed SP3 and SP4. I even clean-reinstalled once or twice. (Course, I did that with RedHat, too.) But it's not the same. I didn't have to muck with the stupid thing. I didn't have to edit any Makefiles or run any configurator telling it what services or hardware I might conceivably need in the future. I double-click the installer and go. When I get new hardware, I install it. Period. Any kernel changes are taken care of by the MS installer, and any hardware dependencies are taken care of by the hardware vendor. I don't have to beat my head against the internals of the system. That's the difference.
I'm not saying at all that I prefer Windows NT to Linux. In fact, my NT box is running Red Hat right now. I'm just saying there are some things you cannot realistically claim Linux does as well as Windows, and kernel/devices/support/upgrades is one of them.
NT drivers are modular. You install them when you install the hardware if they're not part of the stardard distribution. No muss, no fuss, no need to compile in support for new drivers, and no need to load drivers for hardware you don't have! The still take up disk, but not RAM.
You can't take away the GUI, but believe me, NT does not carry around hardware baggage like Unix. There's no bogus/dev directory with precompiled hooks for every device or port you can imagine having.
NT may (cough) not be the world's best operating system (cough cough), but it totally kicks butt over Unix when it comes to adding new physical devices. Assuming a driver exists, which goes both ways. Its autodetect is just as good as RedHat's, its drivers are more modular, and you never have to fsck with the kernel source code or command line options.
for my Initio SCSI card. Show me how to access this fairly common SCSI card with any standard (yea and verily, module-enabled) kernel. (No fair using the Initio-supplied kernel, it doesn't support Ethernet.) I couldn't. It installed just fine on my NT system, though. No kernel support needed.
It also installed just fine on my Red Hat system - after I manually edited the kernel source code! This goes straight to the heart of my gripe against Linux as a mainstream OS.
Oh, yeah. That's the thing you have to set to '3' to make the VCR work. But who in their right mind would watch TV when they could be writing code?;-)
Yes, I am kidding. I code at work. At home I recompile my kernel and swear a lot. Hmm... maybe I should schedule my kernel compiles for when Futurama is on. (How's that for coming back on-topic?)
P.S. Juliet, love the.sig - I think you're trying to provoke a riot!
I've had my Windows NT machine for over two years and never needed to rebuild the kernel. Not even when I added new SCSI and Ethernet cards. Not even when I applied major OS patches. Ditto for my Macs. Can your Linux box do that?
The need for end users to mess with the kernel - ever - is one of Linux' biggest barriers to mainstream acceptance.
The iMac is more than fast enough
on
iMac Linux
·
· Score: 2
The only reason you Linuxers need fast CPUs at all is for recompiling your kernel every night, or for playing those unaccelerated X-windows games.
I have two 68030-based machines that are plenty fast enough for 80% of what I do, one Mac and one BSD box. I have a PPC 601 and a P5-90 that are plenty fast enough in MacOS or Windows (resp.) for 95% of what I do. Oh, yeah, they both run Linux in alternate HD partitions. Linux is fast enough too.
If an iMac isn't fast enough for 100% of your computing needs, my friend, you are not sane.
Apple absolutely knows what open source is all about. If you want a good Apple open source OS, look no further than MkLinux. The OS X/Darwin debacle has nothing to do with open source and everything to do with jumping on the current buzzword bandwagon. It's a boardroom bullet point, nothing more. The restrictive license is to ensure that nobody does anything meaningful with the code. Not that the code is all that useful anyway. BSD source we have aplenty, and who wants a hacked up Mach 2.5, when you can get Mach 3 in MkLinux? That kernel is a carryover from NeXT, and is old news. Don't get upset that Apple is issuing encumbered source code. Just pretend they never issued any source code at all and get on with life.
You people are not getting my point. When I said politicians are more alike than different, I really meant it. That means: 1. It doesn't matter whether they are liberal or conservative. Most politicians of any party (except possibly Libertarian) want to take your money and give it to someone else. You will not be able to persuade me otherwise, so don't try. If you say "Republican", I will laugh out loud. 2. Most politicians, "a" or "b", are good little party players. Who stand by their man, right or wrong. Who vote in blocks, right or wrong. Who vote the way their party boss tells them to. 3. For every politician of flavor "a" who has ever toed the party line, flip-flopped, lied, taken a bribe, cheated on their SO, gotten sexual favors while at work, broken campaign finance laws, put national security at risk, engaged in back-room dealmaking, raised taxes, wrote pork-barrel legislation, or farted in public, I can name one of flavor "b" who did the same or worse. 4. For every dimension in which "a" is good and "b" is bad, I can both find someone who believes the opposite, and name a different dimension in which you will believe the opposite. Different people also happen to value different dimensions more or less highly. 5. Your impression of what "a" and "b" mean is almost certainly different than mine anyway. 6. Unless you're a better rhetorician than I think, you absolutely will not change my mind, because you cannot convince me that the evidence of my own experience is untrue, and I agree not to try to change yours. Or attack or defend the pack of lying, cheating mongrels you prefer to identify with.:-) 7. It's still more fun to talk about T1 lines.
First of all, there are many dimensions to liberalism and conservitism. Secondly, both sides have their good points and their flaws. Thirdly, it's crap anyway because politicians of both flavors are more alike than different. Fourth, you're not going to change my mind and I won't try to change yours. Fifth and finally, it's more fun to talk about T1 lines.
Given the same sort of usage statistics that let ISPs buy one modem for every 20 customers, times the speed advantage of cable modem, you need to have 120 users on the same local loop before the T1 becomes the better bet. My experience is that 60+ X-windows workstations can share a 10Mbps link just fine. Oh, and cable modem in my area costs 40 times less than a T1. I think I'll take my chances with the cable modem.:-)
First of all, people are sheep. Second, in order not to do what the giant corps want you to do, you need a couple things: a) a choice. b) knowledge.
You have to, first, know enough to recognize that something is being foisted off on you, then know enough about it to know it's bad. Not everything being foisted is bad, after all. Unfortunately, you can't be an expert on everything, and sometimes you just have to trust someone else's claims. (No, I have no evidence to back that statement up.)
Once you see something you recognize as bad, you may not have a choice to avoid it. It may be the case that all providers share the undesirable trait. They may even have persuaded the government to mandate it, or had it mandated over their objections.
Some things you can just decide to do without. Some you can't. Some situations are simply inflexible in their requirements. Sometimes you can't afford the preferable choice. Sometimes public funding takes the decision out of your hands.
There are risks to rolling your own solutions, too. I used to grow vegetables in a vacant lot by my house. I have no idea what used to be on that land, or what chemicals are in that soil.
The implication should be clear that the more you know, the less likely you are to get screwed. You also have to be motivated to sometimes make an extra effort, or pay more, or do without, and sometimes that still isn't enough.
Now think about what percentage of people are knowledgeable (not smart) AND motivated AND empowered AND willing to sacrifice. Everyone else is lambs to the slaughter. Is it any wonder that the giant corporations do nearly anything they want, and we do what they want us to do?
Great: 32+ general registers, 32 more FP, plus some special-purpose Good: 16+ general registers, plus FP or special-purpose Fair: 16 general registers, stack or accumulator architecture Totally sucky: Sharing the same register bank by mutually exclusive functional units
Slashdot Mirror
...into the Linux kernel? That's exactly what I am objecting to.
Yes, although I've had Linux for nearly two years, I'm still fairly new to it. So perhaps I could us e a pointer on LILO. In order to use two Ethernet cards of the same type, I have to specify the IRQ and base address of each on the boot: command line. I tried to put two "ether=" statements in lilo.conf, but it rejects those. I tried to use an "append=" statement with both "ether" parameters in it, but it seems to have no effect, either writing to my hard drive boot sector or to the boot floppy. (I do love having the floppy immediately access the HD based kernel, though.) Any idea what I'm doing wrong?
:-)
Regarding DLLs: You are correct, but it's not just apps; most of the OS is implemented in DLLs, too. Get yourself a white-chocolate-macadamia-nut cookie.
So here's my point about modularity: In NT, you don't have to compile hooks into the kernel for every new card that comes down the line. I bought a SCSI card that didn't even exist when my NT kernel was installed, and just plugged the driver right in and rebooted.
In Linux, I had to: 1. Install GCC 2. Install the kernel source (making sure to get the proper version) 3. Download the driver source 4. Patch the kernel sources and makefiles 5. Build the new kernel 6. Install the new kernel 7. Edit lilo.conf and rerun lilo 8. Reboot -- all this just because I used a new kind of SCSI card! Furthermore, I have to repeat this process _EVERY TIME_ I want to upgrade or patch my kernel, since the driver is not part of the standard distro! Bleagh.
That's the essence of free software, isn't it? As a matter of fact, the makers of the SCSI card in question are working to get support built into some distributions (don't ask me why they don't talk to the kernel team). My whole point way back at the beginning of this thread was, SOME operating systems let you add new hardware species without compiling device specific module hooks into the kernel. Like Windows and MacOS. That's all.
windows 95 and NT are best installed in one big partition. No-brainer. No documentation.
For that matter, that was how one source wanted me to do Red Hat, too. Oh, well.
ouch.
Thanks. Do I sound like a Microserf? Read my other notes in this thread and see if you still think Linux' magic module support will help me.
What the do you mean "NT can't do that"? The entire system is modules! What do you think a DLL is?
Finally, you should NOT have to type command line options to boot the OS! Read up on "Multiple ethernet cards" and see if you don't agree with me.
I'm not arguing against the benefits of being able to performance-tune your kernel (although I think it would be better to do that through user-selectable parameters). I'm saying Linux' inability to accept new hardware without explicit kernel support is a crucial flaw. Please refer to my other posts in this thread.
So there's the difference: No source code, no comilation, total separation of drivers from kernel.
Yes, I installed SP3 and SP4. I even clean-reinstalled once or twice. (Course, I did that with RedHat, too.) But it's not the same. I didn't have to muck with the stupid thing. I didn't have to edit any Makefiles or run any configurator telling it what services or hardware I might conceivably need in the future. I double-click the installer and go. When I get new hardware, I install it. Period. Any kernel changes are taken care of by the MS installer, and any hardware dependencies are taken care of by the hardware vendor. I don't have to beat my head against the internals of the system. That's the difference.
I'm not saying at all that I prefer Windows NT to Linux. In fact, my NT box is running Red Hat right now. I'm just saying there are some things you cannot realistically claim Linux does as well as Windows, and kernel/devices/support/upgrades is one of them.
You can't take away the GUI, but believe me, NT does not carry around hardware baggage like Unix. There's no bogus /dev directory with precompiled hooks for every device or port you can imagine having.
NT may (cough) not be the world's best operating system (cough cough), but it totally kicks butt over Unix when it comes to adding new physical devices. Assuming a driver exists, which goes both ways. Its autodetect is just as good as RedHat's, its drivers are more modular, and you never have to fsck with the kernel source code or command line options.
It also installed just fine on my Red Hat system - after I manually edited the kernel source code! This goes straight to the heart of my gripe against Linux as a mainstream OS.
But then, like I said, NT doesn't need to! Is a shame about not being able to rlogin, though.
Oh, yeah. That's the thing you have to set to '3' to make the VCR work. But who in their right mind would watch TV when they could be writing code? ;-)
.sig - I think you're trying to provoke a riot!
Yes, I am kidding. I code at work. At home I recompile my kernel and swear a lot. Hmm... maybe I should schedule my kernel compiles for when Futurama is on. (How's that for coming back on-topic?)
P.S. Juliet, love the
I've had my Windows NT machine for over two years and never needed to rebuild the kernel. Not even when I added new SCSI and Ethernet cards. Not even when I applied major OS patches. Ditto for my Macs. Can your Linux box do that?
The need for end users to mess with the kernel - ever - is one of Linux' biggest barriers to mainstream acceptance.
The only reason you Linuxers need fast CPUs at all is for recompiling your kernel every night, or for playing those unaccelerated X-windows games.
I have two 68030-based machines that are plenty fast enough for 80% of what I do, one Mac and one BSD box. I have a PPC 601 and a P5-90 that are plenty fast enough in MacOS or Windows (resp.) for 95% of what I do. Oh, yeah, they both run Linux in alternate HD partitions. Linux is fast enough too.
If an iMac isn't fast enough for 100% of your computing needs, my friend, you are not sane.
I believe the previous poster was referring to universities with standards. ;-)
Apple absolutely knows what open source is all about. If you want a good Apple open source OS, look no further than MkLinux.
The OS X/Darwin debacle has nothing to do with open source and everything to do with jumping on the current buzzword bandwagon. It's a boardroom bullet point, nothing more. The restrictive license is to ensure that nobody does anything meaningful with the code.
Not that the code is all that useful anyway. BSD source we have aplenty, and who wants a hacked up Mach 2.5, when you can get Mach 3 in MkLinux? That kernel is a carryover from NeXT, and is old news.
Don't get upset that Apple is issuing encumbered source code. Just pretend they never issued any source code at all and get on with life.
You people are not getting my point. When I said politicians are more alike than different, I really meant it. That means: :-)
1. It doesn't matter whether they are liberal or conservative. Most politicians of any party (except possibly Libertarian) want to take your money and give it to someone else. You will not be able to persuade me otherwise, so don't try. If you say "Republican", I will laugh out loud.
2. Most politicians, "a" or "b", are good little party players. Who stand by their man, right or wrong. Who vote in blocks, right or wrong. Who vote the way their party boss tells them to.
3. For every politician of flavor "a" who has ever toed the party line, flip-flopped, lied, taken a bribe, cheated on their SO, gotten sexual favors while at work, broken campaign finance laws, put national security at risk, engaged in back-room dealmaking, raised taxes, wrote pork-barrel legislation, or farted in public, I can name one of flavor "b" who did the same or worse.
4. For every dimension in which "a" is good and "b" is bad, I can both find someone who believes the opposite, and name a different dimension in which you will believe the opposite. Different people also happen to value different dimensions more or less highly.
5. Your impression of what "a" and "b" mean is almost certainly different than mine anyway.
6. Unless you're a better rhetorician than I think, you absolutely will not change my mind, because you cannot convince me that the evidence of my own experience is untrue, and I agree not to try to change yours. Or attack or defend the pack of lying, cheating mongrels you prefer to identify with.
7. It's still more fun to talk about T1 lines.
First of all, there are many dimensions to liberalism and conservitism. Secondly, both sides have their good points and their flaws. Thirdly, it's crap anyway because politicians of both flavors are more alike than different. Fourth, you're not going to change my mind and I won't try to change yours. Fifth and finally, it's more fun to talk about T1 lines.
Given the same sort of usage statistics that let ISPs buy one modem for every 20 customers, times the speed advantage of cable modem, you need to have 120 users on the same local loop before the T1 becomes the better bet. My experience is that 60+ X-windows workstations can share a 10Mbps link just fine. Oh, and cable modem in my area costs 40 times less than a T1. I think I'll take my chances with the cable modem. :-)
First of all, people are sheep.
Second, in order not to do what the giant corps want you to do, you need a couple things:
a) a choice.
b) knowledge.
You have to, first, know enough to recognize that something is being foisted off on you, then know enough about it to know it's bad. Not everything being foisted is bad, after all. Unfortunately, you can't be an expert on everything, and sometimes you just have to trust someone else's claims. (No, I have no evidence to back that statement up.)
Once you see something you recognize as bad, you may not have a choice to avoid it. It may be the case that all providers share the undesirable trait. They may even have persuaded the government to mandate it, or had it mandated over their objections.
Some things you can just decide to do without. Some you can't. Some situations are simply inflexible in their requirements. Sometimes you can't afford the preferable choice. Sometimes public funding takes the decision out of your hands.
There are risks to rolling your own solutions, too. I used to grow vegetables in a vacant lot by my house. I have no idea what used to be on that land, or what chemicals are in that soil.
The implication should be clear that the more you know, the less likely you are to get screwed. You also have to be motivated to sometimes make an extra effort, or pay more, or do without, and sometimes that still isn't enough.
Now think about what percentage of people are knowledgeable (not smart) AND motivated AND empowered AND willing to sacrifice. Everyone else is lambs to the slaughter. Is it any wonder that the giant corporations do nearly anything they want, and we do what they want us to do?
Your humblest servant,
Weasel Boy
Great: 32+ general registers, 32 more FP, plus some special-purpose Good: 16+ general registers, plus FP or special-purpose Fair: 16 general registers, stack or accumulator architecture Totally sucky: Sharing the same register bank by mutually exclusive functional units
That's what they said about every MacOS release since 7.0, which is more than 15 over the last 8 years, counting the interim releases.
Yep, with only VisiCalc and Lotus123 to lead the way, they were pretty innovative about bringing the spreadsheet to a GUI platform.
Nuff said