I have never had any printing problems with Linux. I have printed with a variety of HP laser/deskjets and Canon (sp?) Colorjet printers. In fact, I have never had any problems printing on a local printer or on a Netware printer.
Sure, some printers/scaners may not have drivers for Linux *yet*, but all the old hardware I have seen is supported by Linux (ie a printer you may already have). If you're going to buy a new printer, its easy enough to browse the web and find if it is supported.
"I run Windows 98 for my workstation (the server and office are Linux). I reboot my Windows box about once a day, or day and a half. I don't have a magical stable copy of windows."
Once every couple of hrs or once every couple of days, it doesn't really matter. Compared to the uptimes of Linux/Unix, the difference is insignificant (I have never crashed or needed to reboot Linux or the HP-UX box I admin, except when I compile the kernel). Its like the difference between 1/X and 2/X where X is some very large number.
true - Win95/98 offers a semi-multiuser desktop, but as you stated, it is semi, or partial. I guess the basis of my argument relies on where the desktop starts and stops. With Win95/98, menus, screen savers, etc are not per-user, so yes, users can clobber other users settings.
I seem to have struck a nerve. Actually, this is something I have thought a lot about. What SuSe is doing, while in not on the same degree of magnitude as what MS *could* do - it is a similar idea.
I don't love RH - I, like Linus, appreciate what they are doing to bring Linux to the masses. I don't think Linux should remain some elitist OS for techie nerds only. Let the whole world be free I say (we have enough privliged peoples in this world). Further, _EVERYTHING_ RH does (like many other distros, but not SuSe) is GPLed - this to me is far more important than how many corps. port products to distro X or distro Y. On the other hand, if a proprietary distro (that is what SuSe and other companies are if they keep portions of the OS non-free) takes the majority of the market and everyone starts porting to only their distro, then we have a potential MS all over again (at least an all GPLed distro can be legally copied and rehashed into a compatible distro).
I just call it like I see it. Maybe it is you who should think twice before posting. Linux is my passion and I've been at it for a couple of years now. I've tried many of the distros - Slackware, Debian, OpenLinux, RedHat, PowerLinux...(on a side note, I've even tried FreeBSD, NetBSD, etc) My Linux/BSD collection is starting to look like that of AOL disks. In all, I have been very happy with RedHat from both a tech. and phylisophical point of view.
I would hardly consider myself an "outsider"
BTW - try posting as a non-AC, it gives your argument more credibility.
Embrace Linux and all the free development that has been done but create an excellent installer that adds much value (extend) and only offer it to those that purchase your distro. If this is true of SuSe, I will never use it. Why, because you are allowing yourself to become dependent on one company - much like MS has done to much of the world. What if MS ported the Win32 API and the Windows shell to Linux, offering all the source for to Linux for free, but charging for their distro and only alowing the port to exist on their distro. I doubt that anyone here would buy it - but it is the same idea.
1) Some wealthy teenager gets caught listening to cell calls.
2) His parents can afford a really great lawer that playes the censorship card to convince the jury that her/his client was just listening to a public broadcast (which he/she was)
3) Cell companies, in a effort to keep public trust, implement encryption. However, because all the noise the US gov. has made about strong encryption - the cell companies just use trivial encryption that is easily broken (ie xor with some constant value, etc).
4) Someone on the net develops and releases code that can decrypt this trivial encryption.
5) Cell companies, again in an effort to keep conversations secret, lobby and get a law passed in Congress prohibiting the unauthorized decryption of "private" data. However, since it is *possible* (but not very usefull) to decrypt a message in your head or on paper - this will have set the precidence for making laws agains certian types of thought.
"Wouldn't you be outraged to find your conversation broadcast?"
All cell phones broadcast conversations - how else could it work. I think a better question would be
"Wouldn't the general public be outraged to learn that cell phones broadcase conversations into public airwaves?"
Cell phones are the equivelent of shouting across a room full of people. The problem is that most people *assume* that they are safe (the salesperson would probably tell you it was secure but most don't even think of this). How many people assume that email is private? should it be illegal for SA to read "private" email that is spooled on their machines?
Sooo many people refuse to give their credit card over shttp but they will on a cell phone. The public needs to understand just how insecure cell phones are and cell companies need to take responsibility to inform customes and take action where needed (ie offer real encryption for those making sesitive phone calls)
I feel this is where Linux/Unix makes the most sense - being true multi-user. In windows, you have to worry about little brother - or whoever, clobbering one of your files. On top of that, users compete for their desktop configuration. With Linux, each user has his/her own home directory/desktop setup - and each is protected from other users via a login/password - it makes perfect sense. In fact, NT requires a user/password to use the computer (however, the separation of users isn't so clean). Linux == multi-user and Windows == single-user. Remember, Linux is not windows and trying to use it like windows limit its power
My personal intro to GNU
on
GNU Inside?
·
· Score: 3
I had no idea what GNU or the FSF was before I started using Linux (I didn't even know it existed). It wasn't until I started reading the COPYING, man, info, source files, etc, etc, that came with my distro (some couple years ago) that I started wondering what GNU was. It didn't take long for me to answer my own question (one search on the net was more than enough). From those days on, I have been a OSS/FSF fanatic. I know there are many others like me out there - for this reason, the FSF owes much to the Linux community (just as the Linux community has felt in debt to the FSF and has promoted the use of FSF utilities). In this sense, the GNU-Linux relationship is just that, a relationship - both complement and need each other. Linux users are probably the biggest supporters of the FSF and Linux wouldn't be as far along w/o the FSF.
That said, I'm getting really tired of this flame war. Why take a chance of slowing the Linux momentum by changing the name - this would hurt both GNU and Linux. The FSF will naturally recieve more donations as Linux grows because Linux is so tied to GNU and companies that want to further Linux (ie IBM, Intel) can do so by funding the FSF.
If anything, RMS should work on his on distro or Linux standard that is based on GNU. (ie GNU/Linux 1.0 complient) - in a sence, something like POSIX. This makes a lot more sense than trying to brand the kernel, which has very little to do with GNU
Re:Might be a compromise, but....
on
GNU Inside?
·
· Score: 1
Assuming of course that most Linux users are anti-microsoft. I don't really consider my self anti-microsoft, a more accurate term would be "Anti-Crap(TM)". I just get annoyed when Word crashes my OS, or it takes 5 sec for the start menu to pop-up. So for me, please make a sticker with a red line through a piece of crap.
until the dominate OS(s) are free (as is free speech and free beer). The problem is this - If you base your income around your OS that you sell (ie MS, Be, etc), you want to maintain control of it. Further, becuase your OS is your cash-cow, you need to make sure that it remains different from other systems so that you can keep give folks a reason to change/stay with your OS (this happend to commercial Unix). Also, if you control the system API, you control who writes software for your OS (much like MS does with the Win32 API). Once your OS complies with the one "OPEN standard", it becomes a comadity (like gas - since when did you choose a gas station based on service or quality of fuel? - there mostly the same) and the OS with the lowest price wins (and you no longer have your cash-cow). So I see the acceptance and adoption of Linux/FreeBSD/NetBSD/etc as a prerequisite to this happening.
I think this was a misunderstanding on their part. Linux does support up to 16 processors but, it doesn't scale well after 4 processors. A proper benchmark should extend out to a point where a definate maximum has occured. In a load vs performace graph this should create a horizontal asymptote. To check for scalability, you need to check that the interval between these asymptotes remains approx. the same as you attempt to scale up. What this means is that this interval will remain consistent for Linux until you try to scale past 4 processors, at which time the intervals will get smaller and thus show that the performace gain for each new processor after 4 becomes less and less.
Re:Let's spend another year on 2.2
on
Linux 2.3.0
·
· Score: 1
I'm sure that patches for 2.2 will continue to get released as development on 2.3 moves forward - this is the way it has worked in the past.
I'm running 2.2.5 and I have no problems (then again, I'm not using Cyrix). My guess would be that something didn't build correctly, you have a hardware problem, or you are using hardware/drivers that are still experimentally supported in Linux (ie alpha or beta code).
The following seems to work best for me: make mrproper; make config; make dep; make clean; make zImage; make modules; make modules_install
I'm all for this as long as the bill requires the use of strong signitures - nothing worse than people thinking it is secure when it might not be.
My only concern is this - How would public keys be managed. This seems to be the achilies heel of PGP/RSA (and other public key systems). For example, lets say I sign a credit transaction with my secret key. They, when Visa comes a knocken' I could just say "That's not my sig. Someone must have created their private key using my name." What could they do about it? probably nothing. Unless...lets say you were required to register your public key with a trusted third party (the gov, bank, etc) and they varified you id first (via SSN/Mother's maden name/Address/etc) before your public key was added to the repo. At this point, others can trust that your sig/public key (which were generated by your private key) really came from you (and not just anyone that may have generated a private key using your name).
DCE and Corba are network computing architectures that use control structures defined by metadata (the IDL/ACF files).
Both DCE and CORBA have defined protocols (DCE uses DCE-RPC and CORBA may use one of many protocols including DCE-RPC and IIOP) that define how data should be transfered (marshled). Further, both have the concept of in/out data (parameters).
Both DCe and CORBA IDL (the metadata) can contain user defined data structures (control structures?). These data structures tell the middleware how data should be sent over the network. For example, with DCE IDL, if you have a structure that contains a field of that same structure type, the DCE runtime assumes that you are sending a NULL terminated linked list. Further, the DCE ACF file can be used to change the way data is sent through the interface (metadata).
Its clear that setting up an OS bench mark for PC's with pre-choosen hardware has too many variables. OS X may support this NIC better than OS Y supportes this NIC...the fun can go on and on. Benchmarks have always had too many variables that lead to misleading results. I propose this: Why not have a benchmark that truly represents a real world situation based on cost/benefit. For example, allow $5K to be spent on hardware and software for each OS configuratoin. Of course, the $5k would have to represend real market values (ie vendors could donate software/hardware, but the market value of these donations would be subtraced from the $5K). This would allow each OS to use the hardware that works best for the drivers avaliable. Sure, OS X may support better hardware that is cheaper than the hardware for OS Y, but isn't that what customers should really be interesed in. Further, if OS Y ran better as a cluster, it doesn't matter as long as you don't exceed the $5K limit. In all, I think this kind of benchmark would be the most accurate and it would be the most benificial to customers that have budgets (that would be everyone).
If you thing that any OS exists w/o bugs your in for a huge let down. Linux has always been very open and honest about issues it may have. Proprietary OS are not open, instead they are very closed and you will probably never see the code or hear the conversations that the developers have. The result - Linux is rock-solid, but the newer features have some well-know issues that will be patched. Proprietary OSes also have issues, but you may not know about them and it may take half a year to get a patch.
Also, notice that kernel 2.0.36 was the 36 release of the 2.0 kernel - 2.0.0 had its share of problems. Likewise, the same process will probably happen with the 2.2 kernel as more releases come. For example, I'm using 2.2.5 and it seems to be a good build.
I read an interview of Linus recently (sorry, cant remember where) in which he talked about Linux and micro-kernels. Basically, Linus feels that micro-kernels are more for marketing than anything. A while back it was the big rage and all new OS's needed to have micro-kernel as one of the bullet-features. Now we are finding that micro-kernels generally spend much too long doing IPC to be usefull unless you are using a large number of CPU's (ie more that 4 - in which case sync is much easier that a monolithic kernel). Linux seems to have hit a nice middle ground (for now) in that all the machine dependent code is abstracted out (into the arch directory of kernel build tree) and the bulk of the kernel is loadable modules that are accessed through a well known interface (much like a micro-kernel communicates with other processes to handle system tasks). This results a very modular kernel that runs very fast. The only draw back to all this is scaling SMP server to many CPUs (you have to sync one large code base on many CPU's). I expect that we'll see either (as Linus sugests) a special build for Linux that improves performace on SMP (but would hurt performace on machines with less than 4 CPUs), or a more mature mkLinux that scales to many CPU's better.
The key word here is "critical" A critical application would be something along the lines of a database backend used to track purchase orders. Anyone trying to do something such as this on a MS OS should be black-listed because they don't know what they are doing.
The only reason non-computer-literate folks can use MS is because it comes pre-installed (like the Mac, which also happens to be used by non-techs). Linux can already be purchased pre-installed and KDE/GNOME/??? can easily be used by anyone. The installation and config aspects of Linux should soon be advanced enough for almost anyone to do. This idea that Linux is too hard and that it will be too expensive to retrain people is just MS FUD.
Sorry, I don't mean to be a nag - but threads are a good thing and often make more sense than forking a process. The difference is that on Unix/Linux threads behave just like another process except that they share the heap with the parent thread (ie the child thread may have been called using clone() with flags that tell the cloned process to use the same memory space but different stack - much faster than IPC). This means that UNIX/Linux may have many hundreds of threads in a single process (many DCE services do this to lessen the connection and init time as multiple requests arive). However, windows threads usually never exceed 32 for performace and sync problems.
this is the classic case of converse error - aka guilt by association. Just as not all geeks kill class mates, not all guns are used to kill people. If these parents took any interest in their kids lives they could have easily prevented this. You can pass law after law but anyone intent on killing will - you cannot stop it. Before there were guns, people poisoned other poeple, or they used a bow and arrow, or a club or,...you get the point. For crying out loud, these kids were making pipe bombs that they probably learned how to make on the net - should we outlaw the net? Of course not, who in their right mind wants more of the gov in their life?
Not school itself, rather the social contexts. I felt like the system was setup to favor those that looked and participated in a particular way. I had a few good freinds, but like 90% of those in high school, I wasn't popular and it was hard to deal with. I say "Get Over It"(TM) - it made me a better person. I can look now and see that almost all of those that were popular in high school have gone nowhere - they partied their life away. Likewise, surfing the net or playing games all day is not a good thing and will also take you nowhere (in moderation Ok, but not all the time, as some of us like to joke about). I don't see the value of gaming/surfing 24x7 (the exception would be communication skills learned via chat rooms, email, etc). You would be better off reading a book or holding down a part-time job (a good work ethic is so valuable in college and in the workplace). So in this sence, I agree with the parents out their that are trying to limit the amount of time spent on the net or gaming. These parents are helping the situation not because the net or games (or music, etc) corupt kids, but because they are getting involved in their kids lives.
At one time SCO was an offshoot from MS but for some reason or another (think it had to do with a deal MS made with ???? to not get into the UNIX market - I may be wrong about this) it no longer is. In fact, SCO just won a lengthy court battle with MS that will alow them to remove all MS code from their kernel. so really, SCO has severed all ties with MS.
Slashdot Mirror
I have never had any printing problems with Linux. I have printed with a variety of HP laser/deskjets and Canon (sp?) Colorjet printers. In fact, I have never had any problems printing on a local printer or on a Netware printer.
Sure, some printers/scaners may not have drivers for Linux *yet*, but all the old hardware I have seen is supported by Linux (ie a printer you may already have). If you're going to buy a new printer, its easy enough to browse the web and find if it is supported.
"I run Windows 98 for my workstation (the server and office are Linux). I reboot my Windows box about once a day, or day and a half. I don't have a magical stable copy of windows."
Once every couple of hrs or once every couple of days, it doesn't really matter. Compared to the uptimes of Linux/Unix, the difference is insignificant (I have never crashed or needed to reboot Linux or the HP-UX box I admin, except when I compile the kernel). Its like the difference between 1/X and 2/X where X is some very large number.
true - Win95/98 offers a semi-multiuser desktop, but as you stated, it is semi, or partial. I guess the basis of my argument relies on where the desktop starts and stops. With Win95/98, menus, screen savers, etc are not per-user, so yes, users can clobber other users settings.
I seem to have struck a nerve. Actually, this is something I have thought a lot about. What SuSe is doing, while in not on the same degree of magnitude as what MS *could* do - it is a similar idea.
I don't love RH - I, like Linus, appreciate what they are doing to bring Linux to the masses. I don't think Linux should remain some elitist OS for techie nerds only. Let the whole world be free I say (we have enough privliged peoples in this world). Further, _EVERYTHING_ RH does (like many other distros, but not SuSe) is GPLed - this to me is far more important than how many corps. port products to distro X or distro Y. On the other hand, if a proprietary distro (that is what SuSe and other companies are if they keep portions of the OS non-free) takes the majority of the market and everyone starts porting to only their distro, then we have a potential MS all over again (at least an all GPLed distro can be legally copied and rehashed into a compatible distro).
I just call it like I see it. Maybe it is you who should think twice before posting. Linux is my passion and I've been at it for a couple of years now. I've tried many of the distros - Slackware, Debian, OpenLinux, RedHat, PowerLinux...(on a side note, I've even tried FreeBSD, NetBSD, etc) My Linux/BSD collection is starting to look like that of AOL disks. In all, I have been very happy with RedHat from both a tech. and phylisophical point of view.
I would hardly consider myself an "outsider"
BTW - try posting as a non-AC, it gives your argument more credibility.
Embrace Linux and all the free development that has been done but create an excellent installer that adds much value (extend) and only offer it to those that purchase your distro. If this is true of SuSe, I will never use it. Why, because you are allowing yourself to become dependent on one company - much like MS has done to much of the world. What if MS ported the Win32 API and the Windows shell to Linux, offering all the source for to Linux for free, but charging for their distro and only alowing the port to exist on their distro. I doubt that anyone here would buy it - but it is the same idea.
just my $0.02
Don't forget, anyone can make copies of RH and sell em. The US$79 in only for the "Official" RH from RH.
I got my RH6 for US$12 (including S&H). Or, if you are so lucky - download it from the net.
So far I've no complaints w/ my RH 6.0 running on my Tochiba labtop (Satellite 2535CDS)
1) Some wealthy teenager gets caught listening to cell calls.
2) His parents can afford a really great lawer that playes the censorship card to convince the jury that her/his client was just listening to a public broadcast (which he/she was)
3) Cell companies, in a effort to keep public trust, implement encryption. However, because all the noise the US gov. has made about strong encryption - the cell companies just use trivial encryption that is easily broken (ie xor with some constant value, etc).
4) Someone on the net develops and releases code that can decrypt this trivial encryption.
5) Cell companies, again in an effort to keep conversations secret, lobby and get a law passed in Congress prohibiting the unauthorized decryption of "private" data. However, since it is *possible* (but not very usefull) to decrypt a message in your head or on paper - this will have set the precidence for making laws agains certian types of thought.
fun
I can't believe no one noticed this:
"Wouldn't you be outraged to find your conversation broadcast?"
All cell phones broadcast conversations - how else could it work. I think a better question would be
"Wouldn't the general public be outraged to learn that cell phones broadcase conversations into public airwaves?"
Cell phones are the equivelent of shouting across a room full of people. The problem is that most people *assume* that they are safe (the salesperson would probably tell you it was secure but most don't even think of this). How many people assume that email is private? should it be illegal for SA to read "private" email that is spooled on their machines?
Sooo many people refuse to give their credit card over shttp but they will on a cell phone. The public needs to understand just how insecure cell phones are and cell companies need to take responsibility to inform customes and take action where needed (ie offer real encryption for those making sesitive phone calls)
sorry bout the rant
I feel this is where Linux/Unix makes the most sense - being true multi-user. In windows, you have to worry about little brother - or whoever, clobbering one of your files. On top of that, users compete for their desktop configuration. With Linux, each user has his/her own home directory/desktop setup - and each is protected from other users via a login/password - it makes perfect sense. In fact, NT requires a user/password to use the computer (however, the separation of users isn't so clean). Linux == multi-user and Windows == single-user. Remember, Linux is not windows and trying to use it like windows limit its power
I had no idea what GNU or the FSF was before I started using Linux (I didn't even know it existed). It wasn't until I started reading the COPYING, man, info, source files, etc, etc, that came with my distro (some couple years ago) that I started wondering what GNU was. It didn't take long for me to answer my own question (one search on the net was more than enough). From those days on, I have been a OSS/FSF fanatic. I know there are many others like me out there - for this reason, the FSF owes much to the Linux community (just as the Linux community has felt in debt to the FSF and has promoted the use of FSF utilities). In this sense, the GNU-Linux relationship is just that, a relationship - both complement and need each other. Linux users are probably the biggest supporters of the FSF and Linux wouldn't be as far along w/o the FSF.
That said, I'm getting really tired of this flame war. Why take a chance of slowing the Linux momentum by changing the name - this would hurt both GNU and Linux. The FSF will naturally recieve more donations as Linux grows because Linux is so tied to GNU and companies that want to further Linux (ie IBM, Intel) can do so by funding the FSF.
If anything, RMS should work on his on distro or Linux standard that is based on GNU. (ie GNU/Linux 1.0 complient) - in a sence, something like POSIX. This makes a lot more sense than trying to brand the kernel, which has very little to do with GNU
Assuming of course that most Linux users are anti-microsoft. I don't really consider my self anti-microsoft, a more accurate term would be "Anti-Crap(TM)". I just get annoyed when Word crashes my OS, or it takes 5 sec for the start menu to pop-up. So for me, please make a sticker with a red line through a piece of crap.
until the dominate OS(s) are free (as is free speech and free beer). The problem is this - If you base your income around your OS that you sell (ie MS, Be, etc), you want to maintain control of it. Further, becuase your OS is your cash-cow, you need to make sure that it remains different from other systems so that you can keep give folks a reason to change/stay with your OS (this happend to commercial Unix). Also, if you control the system API, you control who writes software for your OS (much like MS does with the Win32 API). Once your OS complies with the one "OPEN standard", it becomes a comadity (like gas - since when did you choose a gas station based on service or quality of fuel? - there mostly the same) and the OS with the lowest price wins (and you no longer have your cash-cow). So I see the acceptance and adoption of Linux/FreeBSD/NetBSD/etc as a prerequisite to this happening.
I noticed that on their web page they have asked folks not to use MesaGL for legal reasons. How about MesaGPL :)
I think this was a misunderstanding on their part. Linux does support up to 16 processors but, it doesn't scale well after 4 processors. A proper benchmark should extend out to a point where a definate maximum has occured. In a load vs performace graph this should create a horizontal asymptote. To check for scalability, you need to check that the interval between these asymptotes remains approx. the same as you attempt to scale up. What this means is that this interval will remain consistent for Linux until you try to scale past 4 processors, at which time the intervals will get smaller and thus show that the performace gain for each new processor after 4 becomes less and less.
I'm sure that patches for 2.2 will continue to get released as development on 2.3 moves forward - this is the way it has worked in the past.
I'm running 2.2.5 and I have no problems (then again, I'm not using Cyrix). My guess would be that something didn't build correctly, you have a hardware problem, or you are using hardware/drivers that are still experimentally supported in Linux (ie alpha or beta code).
The following seems to work best for me: make mrproper; make config; make dep; make clean; make zImage; make modules; make modules_install
I'm all for this as long as the bill requires the use of strong signitures - nothing worse than people thinking it is secure when it might not be.
:)
My only concern is this - How would public keys be managed. This seems to be the achilies heel of PGP/RSA (and other public key systems). For example, lets say I sign a credit transaction with my secret key. They, when Visa comes a knocken' I could just say "That's not my sig. Someone must have created their private key using my name." What could they do about it? probably nothing. Unless...lets say you were required to register your public key with a trusted third party (the gov, bank, etc) and they varified you id first (via SSN/Mother's maden name/Address/etc) before your public key was added to the repo. At this point, others can trust that your sig/public key (which were generated by your private key) really came from you (and not just anyone that may have generated a private key using your name).
sorry for the horrid spelling - its friday
DCE and Corba are network computing architectures that use control structures defined by metadata (the IDL/ACF files).
Both DCE and CORBA have defined protocols (DCE uses DCE-RPC and CORBA may use one of many protocols including DCE-RPC and IIOP) that define how data should be transfered (marshled). Further, both have the concept of in/out data (parameters).
Both DCe and CORBA IDL (the metadata) can contain user defined data structures (control structures?). These data structures tell the middleware how data should be sent over the network. For example, with DCE IDL, if you have a structure that contains a field of that same structure type, the DCE runtime assumes that you are sending a NULL terminated linked list. Further, the DCE ACF file can be used to change the way data is sent through the interface (metadata).
Does this work - or am I way off?
Its clear that setting up an OS bench mark for PC's with pre-choosen hardware has too many variables. OS X may support this NIC better than OS Y supportes this NIC...the fun can go on and on. Benchmarks have always had too many variables that lead to misleading results. I propose this: Why not have a benchmark that truly represents a real world situation based on cost/benefit. For example, allow $5K to be spent on hardware and software for each OS configuratoin. Of course, the $5k would have to represend real market values (ie vendors could donate software/hardware, but the market value of these donations would be subtraced from the $5K). This would allow each OS to use the hardware that works best for the drivers avaliable. Sure, OS X may support better hardware that is cheaper than the hardware for OS Y, but isn't that what customers should really be interesed in. Further, if OS Y ran better as a cluster, it doesn't matter as long as you don't exceed the $5K limit. In all, I think this kind of benchmark would be the most accurate and it would be the most benificial to customers that have budgets (that would be everyone).
just my 2 cents
If you thing that any OS exists w/o bugs your in for a huge let down. Linux has always been very open and honest about issues it may have. Proprietary OS are not open, instead they are very closed and you will probably never see the code or hear the conversations that the developers have. The result - Linux is rock-solid, but the newer features have some well-know issues that will be patched. Proprietary OSes also have issues, but you may not know about them and it may take half a year to get a patch.
Also, notice that kernel 2.0.36 was the 36 release of the 2.0 kernel - 2.0.0 had its share of problems. Likewise, the same process will probably happen with the 2.2 kernel as more releases come. For example, I'm using 2.2.5 and it seems to be a good build.
I read an interview of Linus recently (sorry, cant remember where) in which he talked about Linux and micro-kernels. Basically, Linus feels that micro-kernels are more for marketing than anything. A while back it was the big rage and all new OS's needed to have micro-kernel as one of the bullet-features. Now we are finding that micro-kernels generally spend much too long doing IPC to be usefull unless you are using a large number of CPU's (ie more that 4 - in which case sync is much easier that a monolithic kernel). Linux seems to have hit a nice middle ground (for now) in that all the machine dependent code is abstracted out (into the arch directory of kernel build tree) and the bulk of the kernel is loadable modules that are accessed through a well known interface (much like a micro-kernel communicates with other processes to handle system tasks). This results a very modular kernel that runs very fast. The only draw back to all this is scaling SMP server to many CPUs (you have to sync one large code base on many CPU's). I expect that we'll see either (as Linus sugests) a special build for Linux that improves performace on SMP (but would hurt performace on machines with less than 4 CPUs), or a more mature mkLinux that scales to many CPU's better.
just my 2 cents
The key word here is "critical" A critical application would be something along the lines of a database backend used to track purchase orders. Anyone trying to do something such as this on a MS OS should be black-listed because they don't know what they are doing.
The only reason non-computer-literate folks can use MS is because it comes pre-installed (like the Mac, which also happens to be used by non-techs). Linux can already be purchased pre-installed and KDE/GNOME/??? can easily be used by anyone. The installation and config aspects of Linux should soon be advanced enough for almost anyone to do. This idea that Linux is too hard and that it will be too expensive to retrain people is just MS FUD.
Sorry, I don't mean to be a nag - but threads are a good thing and often make more sense than forking a process. The difference is that on Unix/Linux threads behave just like another process except that they share the heap with the parent thread (ie the child thread may have been called using clone() with flags that tell the cloned process to use the same memory space but different stack - much faster than IPC). This means that UNIX/Linux may have many hundreds of threads in a single process (many DCE services do this to lessen the connection and init time as multiple requests arive). However, windows threads usually never exceed 32 for performace and sync problems.
this is the classic case of converse error - aka guilt by association. Just as not all geeks kill class mates, not all guns are used to kill people. If these parents took any interest in their kids lives they could have easily prevented this. You can pass law after law but anyone intent on killing will - you cannot stop it. Before there were guns, people poisoned other poeple, or they used a bow and arrow, or a club or, ...you get the point. For crying out loud, these kids were making pipe bombs that they probably learned how to make on the net - should we outlaw the net? Of course not, who in their right mind wants more of the gov in their life?
Not school itself, rather the social contexts. I felt like the system was setup to favor those that looked and participated in a particular way. I had a few good freinds, but like 90% of those in high school, I wasn't popular and it was hard to deal with. I say "Get Over It"(TM) - it made me a better person. I can look now and see that almost all of those that were popular in high school have gone nowhere - they partied their life away. Likewise, surfing the net or playing games all day is not a good thing and will also take you nowhere (in moderation Ok, but not all the time, as some of us like to joke about). I don't see the value of gaming/surfing 24x7 (the exception would be communication skills learned via chat rooms, email, etc). You would be better off reading a book or holding down a part-time job (a good work ethic is so valuable in college and in the workplace). So in this sence, I agree with the parents out their that are trying to limit the amount of time spent on the net or gaming. These parents are helping the situation not because the net or games (or music, etc) corupt kids, but because they are getting involved in their kids lives.
At one time SCO was an offshoot from MS but for some reason or another (think it had to do with a deal MS made with ???? to not get into the UNIX market - I may be wrong about this) it no longer is. In fact, SCO just won a lengthy court battle with MS that will alow them to remove all MS code from their kernel. so really, SCO has severed all ties with MS.