The original comment is talking about Windows XP not 2000
No the original comment (mine) said that Windows 2000 didn't support the 3c590 - so I wondered if XP supported the somewhat newer 3c905.
and there are major differences between the 2 operating systems.
Not very major, under the hood. From what I hear (no I haven't tried it), XP mostly amounts to Windows 2000 with a new set of themes, and more bundled multimedia software. Internally it is (I believe) known as NT 5.1. But hey, what do I know, I've been steering clear for licensing reasons.
...but 2.4 still includes support for all that legacy stuff.
Not quite all. Some drivers fall into disrepair and a few no longer even compile - because apparently nobody still has the hardware or cares.
A good example is the xd driver. It's for a PC-XT hard disk controller - that is, pre-IDE/ATA. Someone reported that it no longer worked (I think in the early 2.3 days, could be wrong) and I remember Linus saying "If you haven't upgraded your hardware in 10 years, why are you upgrading your kernel?" The retort was "Because retrocomputing is fun." Someone actually offered to donate an old xd interface card to any developer who would promise to continue to maintain the driver.
A USB floppy might be OK, but I don't want an external box cluttering up the shelves any more than they already are.
Hey, your call, as long as you don't plan to use the floppy very often you can stow it about anywhere. Valid point though.
And digital cameras tend to use Compact Flash or Memory Sticks or Smart Media, which won't work in a bog-standard floppy drive obviously, so you need a real USB floppy drive.
I've seen digital cameras with real floppy drives. I don't remember if they are USB or what, though.
As I said before, you could always find an ATAPI-based Superfloppy somewhere.
As for USB->serial dongles, that's fine for a modem, maybe, but you don't do sub-microsecond latencies over USB. For a clock you need a serial port right on the computer's main bus
Depends on your definition of "totally different". (:
The 905 is a PCI 10/100 card, the 590 is an ISA 10 MBit card.
No, the 509 is an ISA 10MBit card. Specifically we have:
3c509 - ISA 10Mbps - classic "Etherlink III"
3c509B - ISA 10Mbps - PnP version of 509
3c515 - ISA 10/100 - rare
3c529 - MCA 10Mbps - similar to 509
3c579 - EISA 10Mbps - similar to 509
3c590 - PCI 10Mbps - "Etherlink III PCI"
3c595 - PCI 10/100 - similar to 590
3c900 - PCI 10Mbps - "Etherlink XL PCI" - similar to 590
3c905 - PCI 10/100 - similar to 595
3c905B, 905C, 920, 980, etc - evolutionary changes to 905
Linux uses three drivers for all of the above: 3c509.c (also covers the 529 and 579), 3c515.c, and 3c59x.c (covers all the rest).
Should Linux really support a card that's that old?
Drivers are pretty self-contained. The only problem with supporting old hardware is that when you change an API you need to edit all the old driver files. But if you happen to forget one, and it quits working, life goes on - until a user who has the proper hardware complains, at which time the driver is updated.
The model works quite well.
Now occasionally a single driver will keep getting extended until it supports a wide range of similar hardware, and at some point, the developers split it into an "old hw" driver and a "new hw" driver, possibly with some overlap. This happened a long time ago with the NCR 53c8xx driver, and more recently with the Tulip driver.
Perhaps a second mainstream branch should be started, linux-deprecated or something.
It's called "Linux-2.2" or "Linux-2.0". Both are still being maintained, by Alan Cox and David Weinehall respectively.
Don't randomly spout off baseless claims just to sound good. Windows 2000 natively supports the 3c590; I'm running two of them on my cable modem box at home.
It's not baseless. I honestly could not get a 3c590 to work in Windows 2000, recently. The OS couldn't find a driver. So I went to 3com.com - and they didn't have one either. I had to put a different card in (luckily we had quite a few NICs sitting around).
I don't know what is different between your cards and mine. I do remember being surprised that the '590 wasn't supported.
Have you linux guys even given Windows XP a fair shot?
My primary desktop machine, which runs Linux just fine, is a Pentium 166 with 128MB of RAM.
Will Windows XP run OK on this, or would I have to go out and spend money to buy a new computer? (Having already spent money to buy Windows XP.)
Oh yeah, I also have an original 3c905 Ethernet card (not 3c905B). Is that still supported in Windows XP? I ask because Windows 2000 no longer supports the 3c590, which is a similar (but even older) model.
Ingo published a huge scheduler update that looks promising, might be worth checking it out if you have a system under high load that tends to become poky/etc.
Definitely - but you probably won't notice much difference on most machines - his scheduler was intended to address problems particularly with huge systems. A mere 1-CPU or 2-CPU machine isn't going to see the real benefits.
Which isn't to say the patch is worthless on anything less than 4 CPUs - apparently it beats the old scheduler on all benchmarks. But for most of us, scheduling doesn't take a lot of CPU anyway.
Just wondering about the word "extant". I know what it means, but I have never heard anyone speak it. It just sort of sounds wrong. Do you actually use it in conversation?
I do, all the time. It's probably my favorite "not usually used in spoken language" words to use in spoken language. Deprecate would be my second favorite.
The RTC isn't on the ISA bus, it's part of the CPU chipset.
What about the PIC? I thought that was ISA. Then again modern boards have APICs so maybe one could do without a legacy PIC. I don't know nearly enough about the deep magic of IRQ routing to say for sure.
Floppy I don't want to lose, they're too convenient for small things that don't merit burning an entire CD-R for.
I agree - so get a USB floppy drive (usually advertised as a "digital camera"). A bit more expensive than an ISA floppy drive but what the hey. Keep a few around the office and plug one in when you need it, which for most people isn't often.
Alternatively, floppy drives come in SCSI, or they used to (some SGI boxes have them). And who could forget the (ATAPI) Superdrive?
Serial ports, if I don't have one where am I going to connect the GPS unit for the clock?
Agreed, I need serial ports. If you really want to go legacy-free, lots of people will sell you a USB->serial dongle.
brings back the butt-ugly problem, and looks like it's really difficult to add anything inside, though I suppose that's mainly a job for firewire and USB.
...so stick with the blue-n-white case design. VERY VERY easy to work on - mobo comes out on a hinge, ATA cables are neatly routed out of harm's way like a Real Workstation, power supply doesn't block memory slots, etc, etc, etc.
No, I'm not an Apple fanboy... those are just nice cases.
Re:Singularity doesn't exist
on
True Names
·
· Score: 2
No matter how fast a computering speed is, it will never exist the speed of light.
(I assume you meant "it will never exceed the speed of light.)
Um, I don't mean to be patronising (liar!) but you don't seem to understand the meaning of "speed of a computer". Speed refers to how many calculations of some sort a computer can sustain per unit time. This is only distantly related to measuring the amount of linear displacement per unit time.
I mean really. Are you worried about your computer wandering around your desk at 300000000 meters per second?
Well, all forms of property are artificial notions that come from government.
More precisely, they are notions that come from political philosophers like John Locke and are enforced by the government.
The concept of IP is much newer than that of property rights. (Well, I suppose the "trademark" was foreshadowed by wax seals many thousands of years ago, but those were used more as signatures, not as brand names.)
I wish people would get over the idea that it's an act of government benevolence that I'm allowed to have some control over the creative work that I produce.
And I wish people would get over the idea that IP can be thought of exactly the same as real property.
Look - if you want to think of IP as "just like real property", consider what happens when you sell someone a bushel of oranges. You no longer have any control over what that person does with the oranges. He could plant them and grow his own trees, or eat them, or genetically alter their seeds as a research project. By selling the property, you cede absolute control over it to the buyer.
And yet, when you sell your IP, in the form of bound volumes, or CDs, or numbered lithographs, or software, somehow you expect to keep control over what the buyer does with it.
In summary, you want to have your cake and eat it too. You want "property rights" for your IP, but you don't really want it to be treated like regular property.
...Or think of it this way. If you want control over the IP you produce - don't publish it! Keep it on your coffee table for visitors to marvel at, and don't let anyone borrow it.
Actually, I prefer a longer copyright term to help protect the families of artists.
OK, you currently get life plus 70 years. That will allow your great-grandchildren to continue to profit from your work. Work which they had no part in, nor did their parents, nor their grandparents.
Can you explain why you don't think this is long enough?
If you really write the Great American Novel, but its genius is not recognized until 50 years after you're dead... I say that should just be tough luck for your descendents.
I see no reason for any copyright to extend more than 30 years. If you are still relying financially on something you wrote 30 years ago... get a day job already, you're a has-been, not a great artist.
As for your wife - if you were smart you saved and invested while you were making the big bucks for 30 years, so she should have plenty of inheritance anyway.
Will someone please point out to me the Article, Section, and Clause, of the Constitution which guarantees a right to privacy?
It's not in the print version. The right to privacy was added by the Supreme Court in Roe v. Wade back in the seventies. I guess that was the best they could come up with in terms of proving that it was unconstitutional to deny a woman the right to an abortion [at least given the specific circumstances of the case, which I've forgotten]. Somehow that was violating her privacy, you see.
Say what you will about the right to an abortion (I'm anti, FWIW), it should have remained a legislative matter, not a judicial matter. Oh well.
why should users who can support a more secure situation be penalized by those who cannot? this seems like something that would be easy to make an option.
Right - there should not even be a pap_secrets file if MS-CHAP is used. Do you know whether or not smoothwall does this already? They might. I have no idea.
There are some applications where being able to allocate a big chunk of disk space for a buffer is important, and writing zeroes over it would slow things down. Are you sure there aren't any functions for doing so, hidden in the OS?
Unix has a very useful construct known as "sparse files". Almost all Unix filesystems support them, though "non-native" filesystems (like FAT or ISO9660) do not. A sparse file appears to be just like any other file except for certain disk-block-sized "holes". The holes are not written to disk, do not count against disk free space or your disk quota, but in all other respects behave like regular disk blocks. If you read the file you get zeroes where the holes are. If you write to a hole it is "filled in" (of course, if you write less than a full block, the rest of the block is zeroed).
Thus you can have a 30-megabyte file on a 10-megabyte filesystem, where the 30-meg file really only has 8 megs of non-zero content and 22 megs of zero blocks that don't really exist. If you try to write to the whole file, of course, you'll run out of space.
Aside: this was the source of an interesting glitch with Samba. Windows Explorer copies files by creating the destination file the right size first, to make sure there's room for it, then filling it in (and not doing sufficient error checking on the latter part). The Samba developers had to "fix" Samba awhile back to make sure it created a non-sparse file in that situation.
Similar deal happens with memory. You might think allocating memory would give you access to all kinds of potentially juicy stuff left over from the last process to use that memory. You'd be wrong. The OS clears the memory before letting you use it. With many modern processors, it's possible to optimise this, using memory management tricks, so it doesn't cause the performance hit you might expect.
um as far as i know the/etc/shadow is hashes, ONE WAY.
That's for your local system. The issue under discussion is PPP - which is a protocol with a remote system. Some revisions of the protocol, unfortunately, require that you send a password in cleartext. That password has to be stored somewhere - or, presumably, you could have the user type it in when he boots the firewall... but that's rather inconvenient.
Until you can force all ISPs to migrate to schemes that don't require cleartext passwords (MS-CHAP, for example) you can't fix this one.
The fact that people work on the GPL code of sendmail while on the Sendmail, Inc. payroll seems to be a good enough definition of releasing software under the license.
Sendmail is GPL software? News to me. It looks an awful lot like a BSD license from where I'm sitting.
No, I'm not just being pedantic. There is a huge difference. And Sendmail, Inc. takes full advantage of the difference.
Of course, if you're left handed, this won't work and you're screwed. Just give up your computer work and go pick apples for a living, you sinister left-handed person.
I can't decide if this pun [sinister == Latin for "left"] was intended or not. If so - good one, but a little redundant.
Better yet, you can use the emacs sequence M-x C-r 3.. in place of three dots.
You mean `M-3.' or `C-u 3.' - stop giving Emacs a bad name by implying that its key sequences are obtuse. (:
The decision really had me scratching my head until I realized supreme court judges have one of the 1% of jobs that don't require you to do anything as manual as typing or sweeping, much less clean a toilet.
Come on. They all used to be lawyers. Are you saying a lawyer doesn't have to type, take notes or other such manual activities? Or are you saying that you think the members of the Court are such pricks, by virtue of being in high office, that they have forgotten (to the man, to use an un-PC turn of phrase - remember, it was a 9-0 decision) what life was like as a lawyer?
I didn't read the whole decision but I did read the summary - the Court didn't say anything about manual functions being unimportant in life. They also didn't write the ADA - they are merely interpreting it.
And 3rd parties digging through code.. Im not sure I really like that idea in my environment. No thanks, but i'll stick with companies who's stock prices are well into the double-digits.
What exactly is wrong with 3rd parties digging through code? It's not like the original programmers are all going to be still working for the vendor you buy stuff from. For that matter, the software you buy may itself have been developed elsewhere and later purchased by your vendor - effectively making your vendor a "3rd party digging through code".
If you pay a consultant enough money they should be more than happy to accept the burden of digging through someone else's code. If they turn out to be incompetent to maintain / fix that code, it's the same as if they turn out to be incompetent to do anything else you hire them for.
And what do stock prices have to do with anything? You're saying your company never does professional business with any company whose stock price isn't well into the double digits? If you think the stability of a supplier is in doubt, you don't just walk away from the deal - you make sure you have backup suppliers. Why not the same with support contracts?
This discussion boils down to Administration Philosophies, Open Source Zealotism and professionalism on both sides.
You forgot "Closed Source Zealotism" - or "zealotry", I think the word is. You've said it yourself - for some shops open source is taboo, without there necessarily being any concrete rational reasons.
Well, zealotry rarely hurts anyone other than the zealot (in a free society anyway). If a shop wishes to ignore the various advantages to using open-source solutions, hey, it's their budget.
Take down a network without even trying. Gotta love that power.
Indeed, thanks to the design of NetBIOS and the MSRPC protocols for NT domains, it is quite easy to be a very disruptive influence on a network. And thanks to bugs in the NT implementation, misconfiguring Samba can actually take down NT machines! (Yes, that's a denial-of-service security hole. No, Microsoft doesn't care.) Of course, misconfiguring NT machines can take down NT machines as well - but NT's configuration isn't even close to as easy / flexible as Samba's....
Re:Yeah right why not use Novell crap if you love
on
Samba Turns 10
·
· Score: 2
But at least NetWare has protection. Does Linux? Does NT?
Uh, what? What sort of "protection"? The question is kind of vague.
Linux and NT (and all modern OSes) support memory protection, so one process can't access the memory of another process or of the kernel except by explicit arrangement. They also have crash protection, so an unprivileged user process can't (in theory) cause serious harm to the running of the system.
These have been considered standard features for a serious OS for fifteen years at least. The fact that Windows 95/98/Me and Mac OS 9 didn't have them only means that they should not be considered serious OSes. The fact that Netware 4 didn't have them means it was designed for a niche, not general-purpose.
Slashdot Mirror
No the original comment (mine) said that Windows 2000 didn't support the 3c590 - so I wondered if XP supported the somewhat newer 3c905.
Not very major, under the hood. From what I hear (no I haven't tried it), XP mostly amounts to Windows 2000 with a new set of themes, and more bundled multimedia software. Internally it is (I believe) known as NT 5.1. But hey, what do I know, I've been steering clear for licensing reasons.
Not quite all. Some drivers fall into disrepair and a few no longer even compile - because apparently nobody still has the hardware or cares.
A good example is the xd driver. It's for a PC-XT hard disk controller - that is, pre-IDE/ATA. Someone reported that it no longer worked (I think in the early 2.3 days, could be wrong) and I remember Linus saying "If you haven't upgraded your hardware in 10 years, why are you upgrading your kernel?" The retort was "Because retrocomputing is fun." Someone actually offered to donate an old xd interface card to any developer who would promise to continue to maintain the driver.
I have no idea if anyone took him up on it.
Hey, your call, as long as you don't plan to use the floppy very often you can stow it about anywhere. Valid point though.
I've seen digital cameras with real floppy drives. I don't remember if they are USB or what, though.
As I said before, you could always find an ATAPI-based Superfloppy somewhere.
Fair enough.
Depends on your definition of "totally different". (:
No, the 509 is an ISA 10MBit card. Specifically we have:
3c509 - ISA 10Mbps - classic "Etherlink III"
3c509B - ISA 10Mbps - PnP version of 509 3c515 - ISA 10/100 - rare
3c529 - MCA 10Mbps - similar to 509
3c579 - EISA 10Mbps - similar to 509
3c590 - PCI 10Mbps - "Etherlink III PCI"
3c595 - PCI 10/100 - similar to 590
3c900 - PCI 10Mbps - "Etherlink XL PCI" - similar to 590
3c905 - PCI 10/100 - similar to 595
3c905B, 905C, 920, 980, etc - evolutionary changes to 905
Linux uses three drivers for all of the above: 3c509.c (also covers the 529 and 579), 3c515.c, and 3c59x.c (covers all the rest).
Drivers are pretty self-contained. The only problem with supporting old hardware is that when you change an API you need to edit all the old driver files. But if you happen to forget one, and it quits working, life goes on - until a user who has the proper hardware complains, at which time the driver is updated.
The model works quite well.
Now occasionally a single driver will keep getting extended until it supports a wide range of similar hardware, and at some point, the developers split it into an "old hw" driver and a "new hw" driver, possibly with some overlap. This happened a long time ago with the NCR 53c8xx driver, and more recently with the Tulip driver.
It's called "Linux-2.2" or "Linux-2.0". Both are still being maintained, by Alan Cox and David Weinehall respectively.
It's not baseless. I honestly could not get a 3c590 to work in Windows 2000, recently. The OS couldn't find a driver. So I went to 3com.com - and they didn't have one either. I had to put a different card in (luckily we had quite a few NICs sitting around).
I don't know what is different between your cards and mine. I do remember being surprised that the '590 wasn't supported.
My primary desktop machine, which runs Linux just fine, is a Pentium 166 with 128MB of RAM. Will Windows XP run OK on this, or would I have to go out and spend money to buy a new computer? (Having already spent money to buy Windows XP.)
Oh yeah, I also have an original 3c905 Ethernet card (not 3c905B). Is that still supported in Windows XP? I ask because Windows 2000 no longer supports the 3c590, which is a similar (but even older) model.
Which either means the 2.4 drivers are buggy ... or ... the 2.2 drivers aren't reporting your CRC errors.
Definitely - but you probably won't notice much difference on most machines - his scheduler was intended to address problems particularly with huge systems. A mere 1-CPU or 2-CPU machine isn't going to see the real benefits.
Which isn't to say the patch is worthless on anything less than 4 CPUs - apparently it beats the old scheduler on all benchmarks. But for most of us, scheduling doesn't take a lot of CPU anyway.
I do, all the time. It's probably my favorite "not usually used in spoken language" words to use in spoken language. Deprecate would be my second favorite.
What about the PIC? I thought that was ISA. Then again modern boards have APICs so maybe one could do without a legacy PIC. I don't know nearly enough about the deep magic of IRQ routing to say for sure.
I agree - so get a USB floppy drive (usually advertised as a "digital camera"). A bit more expensive than an ISA floppy drive but what the hey. Keep a few around the office and plug one in when you need it, which for most people isn't often.
Alternatively, floppy drives come in SCSI, or they used to (some SGI boxes have them). And who could forget the (ATAPI) Superdrive?
Agreed, I need serial ports. If you really want to go legacy-free, lots of people will sell you a USB->serial dongle.
...so stick with the blue-n-white case design. VERY VERY easy to work on - mobo comes out on a hinge, ATA cables are neatly routed out of harm's way like a Real Workstation, power supply doesn't block memory slots, etc, etc, etc.
No, I'm not an Apple fanboy ... those are just nice cases.
(I assume you meant "it will never exceed the speed of light.)
Um, I don't mean to be patronising (liar!) but you don't seem to understand the meaning of "speed of a computer". Speed refers to how many calculations of some sort a computer can sustain per unit time. This is only distantly related to measuring the amount of linear displacement per unit time.
I mean really. Are you worried about your computer wandering around your desk at 300000000 meters per second?
More precisely, they are notions that come from political philosophers like John Locke and are enforced by the government.
The concept of IP is much newer than that of property rights. (Well, I suppose the "trademark" was foreshadowed by wax seals many thousands of years ago, but those were used more as signatures, not as brand names.)
And I wish people would get over the idea that IP can be thought of exactly the same as real property.
Look - if you want to think of IP as "just like real property", consider what happens when you sell someone a bushel of oranges. You no longer have any control over what that person does with the oranges. He could plant them and grow his own trees, or eat them, or genetically alter their seeds as a research project. By selling the property, you cede absolute control over it to the buyer.
And yet, when you sell your IP, in the form of bound volumes, or CDs, or numbered lithographs, or software, somehow you expect to keep control over what the buyer does with it.
In summary, you want to have your cake and eat it too. You want "property rights" for your IP, but you don't really want it to be treated like regular property.
...Or think of it this way. If you want control over the IP you produce - don't publish it! Keep it on your coffee table for visitors to marvel at, and don't let anyone borrow it.
OK, you currently get life plus 70 years. That will allow your great-grandchildren to continue to profit from your work. Work which they had no part in, nor did their parents, nor their grandparents.
Can you explain why you don't think this is long enough?
If you really write the Great American Novel, but its genius is not recognized until 50 years after you're dead ... I say that should just be tough luck for your descendents.
I see no reason for any copyright to extend more than 30 years. If you are still relying financially on something you wrote 30 years ago ... get a day job already, you're a has-been, not a great artist.
As for your wife - if you were smart you saved and invested while you were making the big bucks for 30 years, so she should have plenty of inheritance anyway.
It's not in the print version. The right to privacy was added by the Supreme Court in Roe v. Wade back in the seventies. I guess that was the best they could come up with in terms of proving that it was unconstitutional to deny a woman the right to an abortion [at least given the specific circumstances of the case, which I've forgotten]. Somehow that was violating her privacy, you see.
Say what you will about the right to an abortion (I'm anti, FWIW), it should have remained a legislative matter, not a judicial matter. Oh well.
Right - there should not even be a pap_secrets file if MS-CHAP is used. Do you know whether or not smoothwall does this already? They might. I have no idea.
Yes, really.
Unix has a very useful construct known as "sparse files". Almost all Unix filesystems support them, though "non-native" filesystems (like FAT or ISO9660) do not. A sparse file appears to be just like any other file except for certain disk-block-sized "holes". The holes are not written to disk, do not count against disk free space or your disk quota, but in all other respects behave like regular disk blocks. If you read the file you get zeroes where the holes are. If you write to a hole it is "filled in" (of course, if you write less than a full block, the rest of the block is zeroed).
Thus you can have a 30-megabyte file on a 10-megabyte filesystem, where the 30-meg file really only has 8 megs of non-zero content and 22 megs of zero blocks that don't really exist. If you try to write to the whole file, of course, you'll run out of space.
Aside: this was the source of an interesting glitch with Samba. Windows Explorer copies files by creating the destination file the right size first, to make sure there's room for it, then filling it in (and not doing sufficient error checking on the latter part). The Samba developers had to "fix" Samba awhile back to make sure it created a non-sparse file in that situation.
Similar deal happens with memory. You might think allocating memory would give you access to all kinds of potentially juicy stuff left over from the last process to use that memory. You'd be wrong. The OS clears the memory before letting you use it. With many modern processors, it's possible to optimise this, using memory management tricks, so it doesn't cause the performance hit you might expect.
That's for your local system. The issue under discussion is PPP - which is a protocol with a remote system. Some revisions of the protocol, unfortunately, require that you send a password in cleartext. That password has to be stored somewhere - or, presumably, you could have the user type it in when he boots the firewall ... but that's rather inconvenient.
Until you can force all ISPs to migrate to schemes that don't require cleartext passwords (MS-CHAP, for example) you can't fix this one.
Sendmail is GPL software? News to me. It looks an awful lot like a BSD license from where I'm sitting.
No, I'm not just being pedantic. There is a huge difference. And Sendmail, Inc. takes full advantage of the difference.
I can't decide if this pun [sinister == Latin for "left"] was intended or not. If so - good one, but a little redundant.
You mean `M-3 .' or `C-u 3 .' - stop giving Emacs a bad name by implying that its key sequences are obtuse. (:
Come on. They all used to be lawyers. Are you saying a lawyer doesn't have to type, take notes or other such manual activities? Or are you saying that you think the members of the Court are such pricks, by virtue of being in high office, that they have forgotten (to the man, to use an un-PC turn of phrase - remember, it was a 9-0 decision) what life was like as a lawyer?
I didn't read the whole decision but I did read the summary - the Court didn't say anything about manual functions being unimportant in life. They also didn't write the ADA - they are merely interpreting it.
What exactly is wrong with 3rd parties digging through code? It's not like the original programmers are all going to be still working for the vendor you buy stuff from. For that matter, the software you buy may itself have been developed elsewhere and later purchased by your vendor - effectively making your vendor a "3rd party digging through code".
If you pay a consultant enough money they should be more than happy to accept the burden of digging through someone else's code. If they turn out to be incompetent to maintain / fix that code, it's the same as if they turn out to be incompetent to do anything else you hire them for.
And what do stock prices have to do with anything? You're saying your company never does professional business with any company whose stock price isn't well into the double digits? If you think the stability of a supplier is in doubt, you don't just walk away from the deal - you make sure you have backup suppliers. Why not the same with support contracts?
You forgot "Closed Source Zealotism" - or "zealotry", I think the word is. You've said it yourself - for some shops open source is taboo, without there necessarily being any concrete rational reasons.
Well, zealotry rarely hurts anyone other than the zealot (in a free society anyway). If a shop wishes to ignore the various advantages to using open-source solutions, hey, it's their budget.
Indeed, thanks to the design of NetBIOS and the MSRPC protocols for NT domains, it is quite easy to be a very disruptive influence on a network. And thanks to bugs in the NT implementation, misconfiguring Samba can actually take down NT machines! (Yes, that's a denial-of-service security hole. No, Microsoft doesn't care.) Of course, misconfiguring NT machines can take down NT machines as well - but NT's configuration isn't even close to as easy / flexible as Samba's....
Uh, what? What sort of "protection"? The question is kind of vague.
Linux and NT (and all modern OSes) support memory protection, so one process can't access the memory of another process or of the kernel except by explicit arrangement. They also have crash protection, so an unprivileged user process can't (in theory) cause serious harm to the running of the system.
These have been considered standard features for a serious OS for fifteen years at least. The fact that Windows 95/98/Me and Mac OS 9 didn't have them only means that they should not be considered serious OSes. The fact that Netware 4 didn't have them means it was designed for a niche, not general-purpose.