Uh, I definitely recall the 'Arab terrorist' hypothesis lasting about a week or so. The militia/Rush Limbaugh frenzy lasted several months - pretty much during the entire trial.
Better still, when Timothy McVeigh killed hundreds in the bombing of the federal building in OK, where were the mobs running around threatening white males of Christian background?
As I recall, they were all on NPR demanding that Rush Limbaugh be held responsible.
Us OS/2 guys always said the same thing about Windows - why wait for Windows95 when OS/2 had all its features, and stability as well?
Probably because the Windows 3.11 / DOS users wanted their programs to continue working, didn't feel like buying a new printer, and didn't want to learn yet another radically different UI.
Backwards compatibility wins big. Pity Linux can't even manage to stay backwards compatible with itself.
It's not immediately obvious to me why 'inventions that result in things that have physical representations' have some sort of a priori goodness that 'inventions that result in things without physical representations' lack.
Fine. There are at least a half-dozen _invention_ patents for paperclips. Alas, I don't have my copy of Petroski's Evolution of Useful Things handy, so I can't quote the exact patent numbers, but in that book, he examines the patent drawings for paperclips, and how various features were added _as distinct inventions_.
It very definitely has bearing on what's being discussed here.
Geez, doesn't *anybody* know the history of patents? There are literally *hundreds* of patents for (e.g.) paperclips. Each patent describes a slightly different implementation of a paperclip. One might examine, for example, patent 494,622 or patent 371,390 - both patents issued for paperclips, issued in 2004 and 1996, respectively.
Similarly, Microsoft has a patent on a slightly different implementation of setuid.
Oh, wait... This is Microsoft, and therefore evil.
Yah, the Republicans have Limbaugh & Fox News. The Democrats are clearly outmatched, what with the support of CBS, NBC, ABC, CNN, MSNBC, PBS, NY Times, LA Times, etc, etc, etc.
If the democrats can't win with that much media support, maybe the democrats should look at their message.
No, an OS Installation image is designed to (gasp) be installed on a variety of different hardware. You need one of those for bulk installs, and, as I've said before, MS has a different set of tools (outside of Sysprep) that does that sort of thing. More importantly, every single communication from Microsoft that you've cited assumes that you're doing bulk installs, and, given that context, what they're saying is true.
You, on the other hand, have overgeneralized what they've told you into places that are clearly false. I know it's false, because I've done what you've claimed to be impossible.
For doing personal backups, you do not need an OS Installation Image. You at worst need a disk image.
NT Backup works tolerably well for doing personal backups. If you need a disk image utility, then Acronis or Drive Image 2002 are perfectly good options.
In any case, it is NOT NTFS that is at issue here. If you install XP/2K on a FAT32 partition, you'll have the same problem.
The issue, if you had actually understood the knowledgebase article, is that XP/2K has per-machine IDs, and that cloning the hard drive *clones the machine IDs*. When you put the cloned image on two new machines, domain servers get confused.
If your machine doesn't need to talk to an NT domain controller, the problems won't arise.
If you need to generate OS installation images (say, for a corporate rollout), Microsoft has a different set of tools that does that. I'm not terribly familiar with these tools, because I've never needed to do such things.
Personally, I've had good results with Drive Image 2002. NTBackup, as I've said before, requires that you have a working installation of Windows before it runs; Drive Image 2002 has a bootable floppy.
You didn't understand the KB article the first time you read it, and you don't understand it now.
The article *clearly* talks about cloning hard disks for distribution (i.e., building a corporate master image for distributing across an organization), and has nothing to do with backing up your hard disk and restoring it *on the same machine*.
NTFS is *not* crippled, and I've restored backups several times.
It's a pain in the butt, because you have to have a running version of NT to run ntbackup, but it works *just fine*.
I have to restart my Linux installation at least once a week. Any non-trivial use of smbfs on a dual-processor machine (i.e., copying more than two files at once) locks SuSE 9.0 and 9.1 up tighter than a drum. I've also had problems copying files with long filenames with punctuation, often resulting in lockups.
Since Linux doesn't seem to have a crash log facility (like Windows), nor does it seem to be able to kick the screen into console mode on an OOPS (like Windows), I have no idea what's screwed up, and so I....
U.S. Robotics (the modem company) was named after Asimov's fictional company - U.S. Robotics and Mechanical Men.
Hey, it was the '70s - I rather doubt that the founders of U.S. Robotics really planned to sell a couple billion dollars worth of modems back at the founding.
Office 97 can open office XP docs. But say you have Office 95. Then you can go to the Office Converter Pack site and download a converter for Office 95 to let you load office XP docs in Office 95.
The office team spends a lot of time worrying about backward and forward compatibility.
Yes, it's called the Single-Instance Store on Windows Server 2003. When you create files on a Single-instance volume, the system creates a hash of the file, and (lazily) merges files with identical hashes. Copy-on-write semantics apply, so if you modify one of the merged files, the file is split.
Not to be too obvious a plug, but CyberGuard makes a dandy little soho firewall - the SG300 that has stateful packet inspection, a basic IDS (pretty much portscan) and everything else you asked for, for only $250 - well under a grand. Not as cheap as a $50 linksys box, but it uses less power and is way quieter than the $40 133 MHz machine.
SnapGear's Lite2 and Lite2+ firewalls have dialup connection. They're a bit more than a Linksys at $199, because they're a much smaller company than Linksys. Also, SnapGear firewalls run embedded Linux, for those who care.
Even if it was Open Source, you probably couldn't compile it, either. I've got a lot of old (pre-ANSI) C code that certainly won't compile on GCC 3.1, at least with the default flag settings. There are a lot of abandoned Open Source projects that are nothing more than a pile-o-bits.
(Ask NASA about the early Voyager tapes. They can't read them any more, because they don't have the tape drives; even if they did, the tapes only make sense under a particular hardware/OS/program combo, which they don't have, and nobody remembers how to recreate.)
As I read the article, it's not that building a system on top of Linux is hard, it's that supporting a system on top of Linux is hard.
Different distributions have different package systems, filesystem organization, runtime libs, etc. I'm not saying that this is necessarily a bad thing, but it makes supporting J. Random LinuxUser a lot harder.
You are aware, of course, that the Federal Motor Voter law requires that anyone with a drivers license can vote, right?
It went into effect in 1993, unsurprisingly enough.
Hence, only citizens can drive, because all drivers are voters, and only citizens can vote.
Thanks, Democrats!
Re:Looks too much like XP
on
Aethera 1.0
·
· Score: 4, Insightful
So why don't people apply the same insight to Microsoft? There's got to be a thousand people who say that "Microsoft doesn't innovate - every single one of their products was based on.
I think the parent has a point. Open Source has to be held to exactly the same standard as everybody else.
Or, for that matter, Jim Henson's denial of the Muppets on Saturday Night Live. (First season, mostly; rarely on second season.) I understand that they've been deleted from the reruns & videotapes, as well.
While RISC executables are significantly larger than CISC executables, instruction caches significantly reduce (by 90-95%) the memory pressure caused by code fetches (for any architecture). At worst, you might need to increase your Icache size by 4-8K or so.
Embedded systems want smaller code, because it reduces the number of ROMs needed to ship; also, hard real-time systems often turn off caching of all sorts so that they get predictable access times.
In this (rather limited) case, having a specialized instruction set (like ARM's Thumb) makes sense.
"Incremental" and "unimportant" patents are, in fact, the backbone of the patent system. Look at, for example, the thousands of patents for paperclips, soda cans, etc. Each one describes a simple, non-obvious change that improves some characteristic of the object in question.
If you've got the ICBMs in the first place, getting the GPS co-ordinates is trivial; send some agents to the desired target and have them wait around a bit.
Most terrorism targets are public areas, by their very nature. (If, for example, Cheyenne Mountain disappeared, most people wouldn't be terribly bothered.)
Slashdot Mirror
Uh, I definitely recall the 'Arab terrorist' hypothesis lasting about a week or so. The militia/Rush Limbaugh frenzy lasted several months - pretty much during the entire trial.
As I recall, they were all on NPR demanding that Rush Limbaugh be held responsible.
Probably because the Windows 3.11 / DOS users wanted their programs to continue working, didn't feel like buying a new printer, and didn't want to learn yet another radically different UI.
Backwards compatibility wins big. Pity Linux can't even manage to stay backwards compatible with itself.
Could you explain, rather than declaim?
It's not immediately obvious to me why 'inventions that result in things that have physical representations' have some sort of a priori goodness that 'inventions that result in things without physical representations' lack.
Fine. There are at least a half-dozen _invention_ patents for paperclips. Alas, I don't have my copy of Petroski's Evolution of Useful Things handy, so I can't quote the exact patent numbers, but in that book, he examines the patent drawings for paperclips, and how various features were added _as distinct inventions_.
It very definitely has bearing on what's being discussed here.
Geez, doesn't *anybody* know the history of patents? There are literally *hundreds* of patents for (e.g.) paperclips. Each patent describes a slightly different implementation of a paperclip. One might examine, for example, patent 494,622 or patent 371,390 - both patents issued for paperclips, issued in 2004 and 1996, respectively.
Similarly, Microsoft has a patent on a slightly different implementation of setuid.
Oh, wait... This is Microsoft, and therefore evil.
It would be nice if universities were actually oriented around giving students critical skills of being able to think on their own.
Unfortunately, what they actually do is train students to parrot back whatever political swill their professors give them.
It was true 20 years ago, when I went to college, and it doesn't appear to have changed any.
Yah, the Republicans have Limbaugh & Fox News. The Democrats are clearly outmatched, what with the support of CBS, NBC, ABC, CNN, MSNBC, PBS, NY Times, LA Times, etc, etc, etc.
If the democrats can't win with that much media support, maybe the democrats should look at their message.
No, an OS Installation image is designed to (gasp) be installed on a variety of different hardware. You need one of those for bulk installs, and, as I've said before, MS has a different set of tools (outside of Sysprep) that does that sort of thing. More importantly, every single communication from Microsoft that you've cited assumes that you're doing bulk installs, and, given that context, what they're saying is true.
You, on the other hand, have overgeneralized what they've told you into places that are clearly false. I know it's false, because I've done what you've claimed to be impossible.
For doing personal backups, you do not need an OS Installation Image. You at worst need a disk image.
NT Backup works tolerably well for doing personal backups. If you need a disk image utility, then Acronis or Drive Image 2002 are perfectly good options.
In any case, it is NOT NTFS that is at issue here. If you install XP/2K on a FAT32 partition, you'll have the same problem.
The issue, if you had actually understood the knowledgebase article, is that XP/2K has per-machine IDs, and that cloning the hard drive *clones the machine IDs*. When you put the cloned image on two new machines, domain servers get confused.
If your machine doesn't need to talk to an NT domain controller, the problems won't arise.
If you need to generate OS installation images (say, for a corporate rollout), Microsoft has a different set of tools that does that. I'm not terribly familiar with these tools, because I've never needed to do such things.
Personally, I've had good results with Drive Image 2002. NTBackup, as I've said before, requires that you have a working installation of Windows before it runs; Drive Image 2002 has a bootable floppy.
You didn't understand the KB article the first time you read it, and you don't understand it now.
The article *clearly* talks about cloning hard disks for distribution (i.e., building a corporate master image for distributing across an organization), and has nothing to do with backing up your hard disk and restoring it *on the same machine*.
NTFS is *not* crippled, and I've restored backups several times.
It's a pain in the butt, because you have to have a running version of NT to run ntbackup, but it works *just fine*.
I have to restart my Linux installation at least once a week. Any non-trivial use of smbfs on a dual-processor machine (i.e., copying more than two files at once) locks SuSE 9.0 and 9.1 up tighter than a drum. I've also had problems copying files with long filenames with punctuation, often resulting in lockups.
....
Since Linux doesn't seem to have a crash log facility (like Windows), nor does it seem to be able to kick the screen into console mode on an OOPS (like Windows), I have no idea what's screwed up, and so I
SHRUG & REBOOT.
U.S. Robotics (the modem company) was named after Asimov's fictional company - U.S. Robotics and Mechanical Men.
Hey, it was the '70s - I rather doubt that the founders of U.S. Robotics really planned to sell a couple billion dollars worth of modems back at the founding.
Office 97 can open office XP docs. But say you have Office 95. Then you can go to the Office Converter Pack site and download a converter for Office 95 to let you load office XP docs in Office 95.
The office team spends a lot of time worrying about backward and forward compatibility.
Yes, it's called the Single-Instance Store on Windows Server 2003. When you create files on a Single-instance volume, the system creates a hash of the file, and (lazily) merges files with identical hashes. Copy-on-write semantics apply, so if you modify one of the merged files, the file is split.
Not to be too obvious a plug, but CyberGuard makes a dandy little soho firewall - the SG300 that has stateful packet inspection, a basic IDS (pretty much portscan) and everything else you asked for, for only $250 - well under a grand. Not as cheap as a $50 linksys box, but it uses less power and is way quieter than the $40 133 MHz machine.
SnapGear's Lite2 and Lite2+ firewalls have dialup connection. They're a bit more than a Linksys at $199, because they're a much smaller company than Linksys. Also, SnapGear firewalls run embedded Linux, for those who care.
Even if it was Open Source, you probably couldn't compile it, either. I've got a lot of old (pre-ANSI) C code that certainly won't compile on GCC 3.1, at least with the default flag settings. There are a lot of abandoned Open Source projects that are nothing more than a pile-o-bits.
(Ask NASA about the early Voyager tapes. They can't read them any more, because they don't have the tape drives; even if they did, the tapes only make sense under a particular hardware/OS/program combo, which they don't have, and nobody remembers how to recreate.)
As I read the article, it's not that building a system on top of Linux is hard, it's that supporting a system on top of Linux is hard.
Different distributions have different package systems, filesystem organization, runtime libs, etc. I'm not saying that this is necessarily a bad thing, but it makes supporting J. Random LinuxUser a lot harder.
You are aware, of course, that the Federal Motor Voter law requires that anyone with a drivers license can vote, right?
It went into effect in 1993, unsurprisingly enough.
Hence, only citizens can drive, because all drivers are voters, and only citizens can vote.
Thanks, Democrats!
So why don't people apply the same insight to Microsoft? There's got to be a thousand people who say that "Microsoft doesn't innovate - every single one of their products was based on .
I think the parent has a point. Open Source has to be held to exactly the same standard as everybody else.
Or, for that matter, Jim Henson's denial of the Muppets on Saturday Night Live. (First season, mostly; rarely on second season.) I understand that they've been deleted from the reruns & videotapes, as well.
While RISC executables are significantly larger than CISC executables, instruction caches significantly reduce (by 90-95%) the memory pressure caused by code fetches (for any architecture). At worst, you might need to increase your Icache size by 4-8K or so.
Embedded systems want smaller code, because it reduces the number of ROMs needed to ship; also, hard real-time systems often turn off caching of all sorts so that they get predictable access times.
In this (rather limited) case, having a specialized instruction set (like ARM's Thumb) makes sense.
"Incremental" and "unimportant" patents are, in fact, the backbone of the patent system. Look at, for example, the thousands of patents for paperclips, soda cans, etc. Each one describes a simple, non-obvious change that improves some characteristic of the object in question.
(Henry Petroski has amply described this in The Evolution of Useful Things.)
If you've got the ICBMs in the first place, getting the GPS co-ordinates is trivial; send some agents to the desired target and have them wait around a bit.
Most terrorism targets are public areas, by their very nature. (If, for example, Cheyenne Mountain disappeared, most people wouldn't be terribly bothered.)