PC's should not be designed to run ANY operating system. Operating systems should be designed to run on PC's. This is a subtle but very important difference.
No. Phone companies (Telcos) are protected by "common carrier" status that means they are not responsible for the information that is being transmitted over their lines. ISP's have asked for the same protection, but so far it has been denied. There was some hope that now some of the big ISP's have Telco status (including Demon, which is now part of Scottish Telecom, otherwise known as Thus), they could claim common carrier status as well, but this appears not to be the case with Internet traffic.
As an aside, Scottish Law and English Law are different, but not in all cases. So although this ruling may not affect Scotland, the outcome might have been the same had it occurred in Scotland. For those who have been knocking Demon over the settlement, please bear in mind they are one of the few ISP's that provide a FULL usenet feed, and are refusing to bow to pressure to censor the groups they carry. Many other ISP's have already given in.
Have they given any timescales for unbundling the local loop? BT have been dragging their heels in the UK, claiming they can't unbundle it any faster. Would be interesting to see how an almost identical situation pans out in Australia.
The situation may have changed, but last I heard OS/2 was not officially dead - IBM had announced no NEW version developments, but had not stopped supporting the current versions. IIRC the scoring and information systems for the winter olympics and wimbledon (tennis) were supplied by IBM and completely OS/2 based. Not bad for a dead system.
In general most electronics are pretty flexible in terms of supply voltages - they have to be, because not all power supplies are equal. Tolerances of +/- 10% are quite normal, and for individual chips the voltage ranges are often much wider.
Additionally, most computer components are designed to interface to other manufacturers, and so the voltages are standardised for this reason.
24th doesn't sound bad to me, it's a start anyway.
From the article they say that clusters max out at 64 machines, limiting their size - but also it's claimed that the cluster acts like a single machine, so my question is, why can't you cluster the clusters to use 4096 machines. Is it simply a case of (lack of) bandwidth linking the machines together?
But he does have a point. It took MS 3 service packs and several hot fix releases to get NT Y2K compliant. First SP4 was the fix, then SP4 with hotfixes, then SP5 (I didn't go for that), then SP6, then SP6a. This is NOT how to go about fixing bugs.
In case you weren't aware, the entire image of a fat white bearded old man wearing red was CREATED by the Coca-Cola company as an advertising campaign (in the 20s or 30s?). Up to that point, father christmas had been portrayed wearing traditional greens and browns. The image took off, and since then everyone uses the "coca-cola" father christmas. So YES they DO own it.
The advantage of RAID is often when reading data, not when writing - reads are served by several disks and can be faster, writes require extra parity information, and are slower
IDE RAID is certainly an option, although not for high end systems (for the same reasons, no elevator seeks etc). Anywhere you want reliability, RAID is useful, and the underlying disk technology is irrelevant. I would rather have IDE RAID 5 than plain IDE, or even plain SCSI, where my data is important. I notice that you can get a SCSI RAID device that internally uses IDE disks (i.e it has a SCSI interface to the computer, but the disks and internal controllers are IDE) - much cheaper than a similar all SCSI system.
Does anyone know why this is USA and Canada only? As a UK resident I would more than happy to help with the beta, but for some strange reason I seem to be excluded. Personally I feel this is an unnecessary restriction, and a bit of a slap in the face for us non American existing customers
Are you seriously suggesting that having extra disks made up for less RAM and processor speed - please bear in mind the second array was using a PARITY striped disk, which means for every write it had to update the parity as well. If you examine your RAID information, striping with parity SLOWS disks. Using arrays generally does not speed up disk throughput, but provides redundancy for failure.
I do not wish to be rude, but it seems clear you do not understand the technology. The advantage in the second machine is that SCSI is designed for multi-tasking (elevator seeks, scatter-gather, etc.) and IDE has NONE of these. There is a reason why a SCSI controllor is more expensive than an IDE one, it has in built intelligence, and effectively creates a storage network on a seperate bus, rather than expecting the host computer to control everything.
I would be suprised if IDE ever surplants SCSI for servers, it just isn't designed for it - case in point, running MS SQL server (6.5) on NT Server 4, which of these two machines do you think performed best:
P2 350MHz, 128Mb RAM, 2 x 6 Gb IDE DMA/33 (striped)
Having had a think about this, I suspect this has much less to do with Open Source, and more to do with Market Share (tm). I believe WinCE is heavily outsold by Palm in the US, and as far as OS licensing goes is under extreme threat from Symbian (Psion)'s EPOC32. In particular, the licensing costs of the two (WinCE and EPOC) are hugely different - something like $50 for WinCE and $5 for EPOC, which makes quite a difference to machines that need to make a profit at $200-$300. This could simply be an attempt to regain momentum by giving away the OS, perhaps with the plan of a paid for upgrade later on.
Given that hardware generations are currently at 18 months, and MS wants to move software generations to 12 months (i.e. a new release of Windows xxxx every year), 30 years of unix could be expressed as 20-30 human generations (400-600 years?). I think if a company has been around that long people would tend to think of it as well established, and reputable.
Sorry, but we have some very firm laws on government procurement in the UK. Socialism does NOT equal corruption - if it did, then right wing dictatorships would be totally above board wouldn't they? When will Americans learn that there are more ways of running a country than being a right-wing republic, and governments are not intrinsically evil just because they are different.
They're different from us, so they're evil. Lets kill them. Is this truly the message of America? I hope not.
I've run into a higher frequency of XFree86 crashes than NT crashes.
I would say on average XFree86 locks up every other month or so on my PC, in comparison a freshly installed NT I have has locked up twice in a week so far.
I don't run XFree on my mail server, and thats another point - at least with Linux I have a choice not to - I have had my NT server at work lock up in user32.dll - effectively preventing any appliactions being started (user domain manager, task manager, etc). On logging out to see if that would clear it, I couldn't log back in (without user32.dll running). I think I have traced the fault to the UPS monitoring software - but the result was a mess. I would say I get at least one blue screen every two months with NT on the server. The linux server has died once in six months unexplainedly.
A microkernel is an architectural design, and has nothing to do with being able to change TCP/IP settings dynamically (which you can do in Windows 2000). The NT kernel, like Mach, was designed specifically to run OS emulation layers (called subsystems in NT, and servers in Mach). These emulation layers, more than one of which may run concurrently, actually provide the system-call interfaces to user applications.
I will have to accept your description here! I was under the impression that part of being a microkernel meant that services were modular and therefore should be able to load an unload cleanly, which NT 4 should be able to do, but seems to choose not to. I was also always told a monolithic system had the services compiled in, and therefore Linux doesn't match that description any more with it's use of modules.
Copyright is automatic, but surely you have to claim it, or make the user aware, otherwise how do they know that copyright applies - should anything that does NOT have copyright state 'Copyright not applicable' to let users know they CAN copy it? I would have thought that unless you express copyright, although you may own it, you aren't enforcing it (a different issue - I may write a program, own the copyright, but choose not to enforce it).
I find XFree86 stable enough - sure it locks up occasionally, but in general it is more stable than similarly configured Windows products. The fact that when X goes down I can restart it without rebooting IS a sizeable comfort - it makes it much quicker to recover for a start. It also means that other stuff I might have running in a console session isn't disturbed. Or if I am running a GUI tool on my mail server, it means it doesn't disturb people picking up e-mail.
The widget set comment is fair comment - it's something that needs sorting out, although that being said Windows apps are not automatiaclly consistent - view the differences between old Borland apps and Microsofts.
Finally, Win32 is not much more than a GUI and kernel, I think you are confusing Microsoft's insistence on programming using their MFC libraries with the fundamental OS. It is very quick and easy to write Windows programs that talk direct to the API (I do, I hate MFC). If people did this, then it would be a matter of either replacing or converting these calls to X equivalents. Still difficult, but much easier than implementing MFC etc.
I don't think I would describe NT as a microkernel, it's not very modular (hence the need to restart it if you change TCP/IP settings).
But why isn't Linux used on the desktop? Could it be due to the (perceived or otherwise) lack of 'standard' desktop applications?
Besides, you are missing the point of the articles - as it points out, at the moment everyone who chooses Linux over Windows xx (and there are some, and that number is likely to grow, even if it's just because it's easier to grow a small market share than a big one) deprives MS of an OS customer and an Apps customer. If the number of people using Linux grows, then the likelyhood of a Linux port grows significantly for purely economic reasons - some people will choose Linux anyway, it's better to make $150 from each of them, then $0. Not everyone who uses Linux now, and in the future, objects to paying for software.
I don't disagree with you - it's not a big market numbers-wise for MS, but you are missing one possible argument - it is a NON-SATURATED market. Existing PC and Mac users are harder to sell with, because the chances are they already have several MS products, and are less likely to want to upgrade. So a new market of 3-5% could be really quite profitable
Surely without a digital signature, or BOTH "original" messages (senders and receivers), they can't prove much - it's trivial to forge an email in your own mailbox - I could simply (manually) create a message on my server, and tell my email program to retrieve messages from that rather than my ISP's server. Result, a message that has all the right headers for whoever I want to set up. Likewise I could do the same to make it appear I had sent a message to someone.
It's about time lawyers and the law was dragged into the 20th century, just as we're about to leave it...
Slashdot Mirror
PC's should not be designed to run ANY operating system. Operating systems should be designed to run on PC's. This is a subtle but very important difference.
No. Phone companies (Telcos) are protected by "common carrier" status that means they are not responsible for the information that is being transmitted over their lines. ISP's have asked for the same protection, but so far it has been denied. There was some hope that now some of the big ISP's have Telco status (including Demon, which is now part of Scottish Telecom, otherwise known as Thus), they could claim common carrier status as well, but this appears not to be the case with Internet traffic.
As an aside, Scottish Law and English Law are different, but not in all cases. So although this ruling may not affect Scotland, the outcome might have been the same had it occurred in Scotland. For those who have been knocking Demon over the settlement, please bear in mind they are one of the few ISP's that provide a FULL usenet feed, and are refusing to bow to pressure to censor the groups they carry. Many other ISP's have already given in.
Have they given any timescales for unbundling the local loop? BT have been dragging their heels in the UK, claiming they can't unbundle it any faster. Would be interesting to see how an almost identical situation pans out in Australia.
The situation may have changed, but last I heard OS/2 was not officially dead - IBM had announced no NEW version developments, but had not stopped supporting the current versions. IIRC the scoring and information systems for the winter olympics and wimbledon (tennis) were supplied by IBM and completely OS/2 based. Not bad for a dead system.
In general most electronics are pretty flexible in terms of supply voltages - they have to be, because not all power supplies are equal. Tolerances of +/- 10% are quite normal, and for individual chips the voltage ranges are often much wider.
Additionally, most computer components are designed to interface to other manufacturers, and so the voltages are standardised for this reason.
I stand corrected. Thank you.
24th doesn't sound bad to me, it's a start anyway.
From the article they say that clusters max out at 64 machines, limiting their size - but also it's claimed that the cluster acts like a single machine, so my question is, why can't you cluster the clusters to use 4096 machines. Is it simply a case of (lack of) bandwidth linking the machines together?
Here in the UK Tesco (a supermarket) are selling a Wharfdale DVD which can be de-regioned. It got a best buy as well in one of the hi-fi mags.
But he does have a point. It took MS 3 service packs and several hot fix releases to get NT Y2K compliant. First SP4 was the fix, then SP4 with hotfixes, then SP5 (I didn't go for that), then SP6, then SP6a. This is NOT how to go about fixing bugs.
In case you weren't aware, the entire image of a fat white bearded old man wearing red was CREATED by the Coca-Cola company as an advertising campaign (in the 20s or 30s?). Up to that point, father christmas had been portrayed wearing traditional greens and browns. The image took off, and since then everyone uses the "coca-cola" father christmas. So YES they DO own it.
The advantage of RAID is often when reading data, not when writing - reads are served by several disks and can be faster, writes require extra parity information, and are slower
IDE RAID is certainly an option, although not for high end systems (for the same reasons, no elevator seeks etc). Anywhere you want reliability, RAID is useful, and the underlying disk technology is irrelevant. I would rather have IDE RAID 5 than plain IDE, or even plain SCSI, where my data is important. I notice that you can get a SCSI RAID device that internally uses IDE disks (i.e it has a SCSI interface to the computer, but the disks and internal controllers are IDE) - much cheaper than a similar all SCSI system.
Does anyone know why this is USA and Canada only? As a UK resident I would more than happy to help with the beta, but for some strange reason I seem to be excluded. Personally I feel this is an unnecessary restriction, and a bit of a slap in the face for us non American existing customers
Are you seriously suggesting that having extra disks made up for less RAM and processor speed - please bear in mind the second array was using a PARITY striped disk, which means for every write it had to update the parity as well. If you examine your RAID information, striping with parity SLOWS disks. Using arrays generally does not speed up disk throughput, but provides redundancy for failure.
I do not wish to be rude, but it seems clear you do not understand the technology. The advantage in the second machine is that SCSI is designed for multi-tasking (elevator seeks, scatter-gather, etc.) and IDE has NONE of these. There is a reason why a SCSI controllor is more expensive than an IDE one, it has in built intelligence, and effectively creates a storage network on a seperate bus, rather than expecting the host computer to control everything.
I would be suprised if IDE ever surplants SCSI for servers, it just isn't designed for it - case in point, running MS SQL server (6.5) on NT Server 4, which of these two machines do you think performed best:
P2 350MHz, 128Mb RAM, 2 x 6 Gb IDE DMA/33 (striped)
Pentium 166MHz, 80Mb RAM, 5 x 2Gb Wide SCSI (striped w/parity)
I'll give you a clue, the faster machine with more RAM ran like a total dog whenever any major database work was done.
Having had a think about this, I suspect this has much less to do with Open Source, and more to do with Market Share (tm). I believe WinCE is heavily outsold by Palm in the US, and as far as OS licensing goes is under extreme threat from Symbian (Psion)'s EPOC32. In particular, the licensing costs of the two (WinCE and EPOC) are hugely different - something like $50 for WinCE and $5 for EPOC, which makes quite a difference to machines that need to make a profit at $200-$300. This could simply be an attempt to regain momentum by giving away the OS, perhaps with the plan of a paid for upgrade later on.
Given that hardware generations are currently at 18 months, and MS wants to move software generations to 12 months (i.e. a new release of Windows xxxx every year), 30 years of unix could be expressed as 20-30 human generations (400-600 years?). I think if a company has been around that long people would tend to think of it as well established, and reputable.
Sorry, but we have some very firm laws on government procurement in the UK. Socialism does NOT equal corruption - if it did, then right wing dictatorships would be totally above board wouldn't they? When will Americans learn that there are more ways of running a country than being a right-wing republic, and governments are not intrinsically evil just because they are different.
They're different from us, so they're evil. Lets kill them. Is this truly the message of America? I hope not.
Inprise is the name of the company formerly known as Borland
As both companies have interests in Linux, they are talking of (or in the process of?) merging.
I think alot of sites ran the story whilst waiting for some sort of reaction from Sony - which is understandable
From one of the reports I've seen, the problem is with a particular game, and might still be a fault with the memory cards
I've run into a higher frequency of XFree86 crashes than NT crashes.
I would say on average XFree86 locks up every other month or so on my PC, in comparison a freshly installed NT I have has locked up twice in a week so far.
I don't run XFree on my mail server, and thats another point - at least with Linux I have a choice not to - I have had my NT server at work lock up in user32.dll - effectively preventing any appliactions being started (user domain manager, task manager, etc). On logging out to see if that would clear it, I couldn't log back in (without user32.dll running). I think I have traced the fault to the UPS monitoring software - but the result was a mess. I would say I get at least one blue screen every two months with NT on the server. The linux server has died once in six months unexplainedly.
A microkernel is an architectural design, and has nothing to do with being able to change TCP/IP settings dynamically (which you can do in Windows 2000). The NT kernel, like Mach, was designed specifically to run OS emulation layers (called subsystems in NT, and servers in Mach). These emulation layers, more than one of which may run concurrently, actually provide the system-call interfaces to user applications.
I will have to accept your description here! I was under the impression that part of being a microkernel meant that services were modular and therefore should be able to load an unload cleanly, which NT 4 should be able to do, but seems to choose not to. I was also always told a monolithic system had the services compiled in, and therefore Linux doesn't match that description any more with it's use of modules.
Copyright is automatic, but surely you have to claim it, or make the user aware, otherwise how do they know that copyright applies - should anything that does NOT have copyright state 'Copyright not applicable' to let users know they CAN copy it? I would have thought that unless you express copyright, although you may own it, you aren't enforcing it (a different issue - I may write a program, own the copyright, but choose not to enforce it).
I find XFree86 stable enough - sure it locks up occasionally, but in general it is more stable than similarly configured Windows products. The fact that when X goes down I can restart it without rebooting IS a sizeable comfort - it makes it much quicker to recover for a start. It also means that other stuff I might have running in a console session isn't disturbed. Or if I am running a GUI tool on my mail server, it means it doesn't disturb people picking up e-mail.
The widget set comment is fair comment - it's something that needs sorting out, although that being said Windows apps are not automatiaclly consistent - view the differences between old Borland apps and Microsofts.
Finally, Win32 is not much more than a GUI and kernel, I think you are confusing Microsoft's insistence on programming using their MFC libraries with the fundamental OS. It is very quick and easy to write Windows programs that talk direct to the API (I do, I hate MFC). If people did this, then it would be a matter of either replacing or converting these calls to X equivalents. Still difficult, but much easier than implementing MFC etc.
I don't think I would describe NT as a microkernel, it's not very modular (hence the need to restart it if you change TCP/IP settings).
But why isn't Linux used on the desktop? Could it be due to the (perceived or otherwise) lack of 'standard' desktop applications?
Besides, you are missing the point of the articles - as it points out, at the moment everyone who chooses Linux over Windows xx (and there are some, and that number is likely to grow, even if it's just because it's easier to grow a small market share than a big one) deprives MS of an OS customer and an Apps customer. If the number of people using Linux grows, then the likelyhood of a Linux port grows significantly for purely economic reasons - some people will choose Linux anyway, it's better to make $150 from each of them, then $0. Not everyone who uses Linux now, and in the future, objects to paying for software.
I don't disagree with you - it's not a big market numbers-wise for MS, but you are missing one possible argument - it is a NON-SATURATED market. Existing PC and Mac users are harder to sell with, because the chances are they already have several MS products, and are less likely to want to upgrade. So a new market of 3-5% could be really quite profitable
Surely without a digital signature, or BOTH "original" messages (senders and receivers), they can't prove much - it's trivial to forge an email in your own mailbox - I could simply (manually) create a message on my server, and tell my email program to retrieve messages from that rather than my ISP's server. Result, a message that has all the right headers for whoever I want to set up. Likewise I could do the same to make it appear I had sent a message to someone.
It's about time lawyers and the law was dragged into the 20th century, just as we're about to leave it...