It actually does make sense if the delay in completing the MSH environment is down to providing command line interfaces to every aspect of the OS, as opposed to completing the shell itself. MS Exchange has a much smaller set of features to provide command-line interfaces to than the entire MS Windows system.
You can get that today, in Window XP 64-bit Edition for x64 (or some name like that), which is based on the NT 5.2 kernel, like Windows Server 2003. The only problem with it is device drivers: if all your hardware is supported by drivers from Microsoft, it's fine. If not, you'll have to wait for the hardware vendors to port the required drivers to 64-bit (some have already, whilst others won't commit to ever doing it).
I think AppleScript is more like the existing Windows Scripting technologies. The difference with MSH is that it's an attempt to merge that sort of thing into an interactive command-line environment, and include every aspect of the system that the GUI can manage. (I don't know if AppleScript covers every aspect of the system that is accessible from the GUI, but I know Windows Scripting doesn't.)
And ten years ago, a new shell would have made sense. Now MS have pretty good remote administration in the graphical shell. Most users are sold on Explorer. Why do they need a shell now?
Unix administrators like managing their systems from a command-line rather than graphical interface, and a lot of them are moving from Unix/Risc to Linux/x86/amd64, instead of to Windows/x86/amd64, as Microsoft would like.
In other words, this new shell isn't so much for current users, and certainly not for desktop users, as it is to attract Unix and Linux administrators to Windows. Microsoft Services for Unix helps, by offering a Unix command line and allowing Unix software to run on the NT kernel, but the problem is that Windows services can't be managed from the Unix command line.
They're trying to create an object-oriented shell, where all system components are accessible as objects. This is naturally much more complex than a shell and command-line tools based on manipulation of text streams, but includes a lot of theoretical advantages, if it works.
In general, I like the 'keep it simple' philosophy of traditional UNIX, but I'll try MSH, when it gets to a more mature stage, before deciding whether or not Microsoft have come up with a better CLI than the simple UNIX model.
Win32 was actually released with Windows NT 3.1, in 1993, which makes it 12 years old. NT development started in 1988 or so, but I don't think the Win32 subsystem was started until about 1990, round about the time Microsoft and IBM parted ways (because Microsoft wanted Win32, instead of OS/2, to be the primary API/subsystem for NT).
The anti-Linux comments from EDS didn't sound rabid to me, just like a sales pitch for Solaris.
The conflicting claims make EDS look a bit stupid, but I don't think most objective people think criticism of Linux 'shows religious bias', even if it's vague and perhaps invalid.
'Port' means the version of the software for a given platform. Opera is very portable (or 'cross-OS capable' as you say), which is why there are so many ports!
What does this mean for Minimo (Mozilla's mini-browser that was funded/supported by Nokia)?
Nokia probably don't want to put all of their eggs into a single basket, as the saying goes. If the difference between the value added by Opera and the price of Opera is greater than the value added by Minimo, minus any funding from Nokia, then Opera is a better choice. If not, Minimo may be the better one.
Opera runs on most PC and PDA/mobile OSes, including Windows Mobile. Nokia, however, don't offer any phones running Windows Mobile, so you're right that the Nokia phones the article refers to run Symbian OS.
The majority of people won't care, as long as they don't get infected.
The article doesn't make clear why there will be up to a month delay. It may mean Microsoft will simply test on the configurations used by American government customers first, which wouldn't really delay patches for the rest of us, so wouldn't be so bad.
If this policy is only for the American government, that bothers me more, but not enough to use an inferior substitute. If it's for all governments, and doesn't delay getting patches to other users, I don't care at all.
Unlike Linus Torvalds, Richard Stallman isn't a good leader or a good representative of an open source project. He wants people to use a strange and ugly-sounding name instead of a nice, simple, pleasant-sounding one for his own ego. What else does calling it 'GNU/Linux' accomplish?
Stallman is also very political, and alienates a lot of people (like me) with his divisive rhetoric. I've nothing against Microsoft, Apple, Sun or anyone else who would like to sell proprietary software. I've nothing against those who would like to give away software either, and would rather have source code, but only if those who wrote it want to give it to me. I therefore find Stallman's 'software war' mentality alien and off-putting.
Linux is just a kernel, and would not exist without GNU, but without Linux, FreeBSD would probably be the dominant open source OS today. Stallman should be grateful that Torvalds promoted his project via the Linux kernel, because it would otherwise have never caught on, and the GPL would probably be far less popular than it is.
When you get a Dell, you get Windows XP neutered Home Edition.
Upgrading a Dell or HP to XP Professional would probably still leave it cheaper than a similar Mac. Some of us also prefer the Windows XP GUI to Mac OS X one, and appreciate all the applications available for Windows XP.
My favourite window manager is ion, but the rest of the GUI on Unix-like systems is pretty random, so Windows XP is best overall for me (although I would really like something like ion on it). I think the Mac OS X command line is a bit better than Interix on XP (mostly because more tools have been ported to it), but either one will work for me.
With lots of pointless eye candy and all the spyware and viruses you can eat.:P
I've used Windows XP since it came out, and I've never had any virus/worm/spyware infections.
Certification is simply a guarantee by IBM and Red Hat that a given application meets a set of requirements which they (IBM an RH) have defined. This allows potential customers to look at a set of certification criteria and, if the criteria meets their needs, trust that the associated list of certified applications will do the same.
IBM want to develop the market in such a way that the value is server applications, and especially services, where IBM are strong, rather than in operating systems or commodity hardware, where IBM have long been hopelessly behind Microsoft, Dell et al.
Partnering with multiple Linux vendors will help prevent any one becoming dominant in the market. This, in turn, will prevent the Linux vendors adding too much value to the operating system, which, in IBM's view, should be a commodity layer for running IBM server applications, supported by IBM services.
And yet, Dell's tech support is in India, whereas IBM's tech support is in the US. Hmm...
I don't know if it's the same thing, but the link in the parent post says this:
Dell has some first-hand experience in the problems that can accompany offshoring.
A year ago the company moved customer support operations for its Optiplex desktops and Latitude notebooks back to the U.S. after corporate customers complained about the quality of service they were receiving from international facilities.
If the HP-UX market is dying, it means the market for HP's expensive PA-RISC hardware is dying, since the two are effectively a single product. This might be a problem if the PA-RISC/HP-UX business is a profitable one.
HP's partership with Intel to produce the Itanium was an attempt to produce a replacement for PA-RISC (which, with Alpha, was always a top performer in the RISC market) with wider market appeal. However, we all know how that's been going...
Yes, it's important to remember that proprietary is not the same thing as closed-source. The software released by Red Hat that is defined by Red Hat, rather than strict implementations of standards defined by bodies like ISO, is proprietary, even if the code is shared. This is because Red Hat have defined it, and can change its definition at any time (and so may any other user of the code).
Another interesting thing is that open source doesn't eliminate the 'network effect', as it's called in economics, which has been central to Microsoft's dominance of the operating systems market. The more people there are using Red Hat Linux, the more valuable it becomes, and the availability of source code doesn't change that. It makes it easier for competitors to interoperate with Red Hat Linux, but if Red Hat Linux had sufficient market share (e.g. similar to Microsoft Windows), it's quite possible that competitors would find it extremely difficult to get the market to accept alternatives to it.
As anyone familiar with the rationale behind organisations like ISO and The Open Group will know, open source and open standards are two different things. Both are important, but, at the end of the day, I think open standards are very much more so.
I see your point, but, architecturally, there are only two Microsoft Windows systems: Windows NT and Windows CE. There used to be a third one, MS-DOS/Windows (Windows 1.0 through Windows Me were variants of this system), but Microsoft stopped developing it about five years ago. All of the modern desktop/server variants of Microsoft Windows are in fact different versions of Windows NT.
Windows CE is a very different thing, but it's a limited system designed for small devices, and so isn't relevant to a discussion of full desktop/server systems like Microsoft Windows (NT), Sun Solaris or Red Hat Linux.
I think the root of all this is that the UNIX System includes the operating system, the shell(s) and a wide array of commands (but no GUI). They're all part of the UNIX System, but not part of the operating system. However, since the operating system is also called UNIX (the UNIX Operating System, as distinct from the UNIX System), there has long been a great deal of confusion about what 'UNIX' is (since it's both!).
Linux is comparable to the UNIX Operating System, not the UNIX System, but, following UNIX tradition, most people refer to both the OS proper and the entire system as simply 'Linux', even if, techically, only the former is. The ugliness of the GNU name (both to the eyes and the ears) may be part of the reason most people don't call it GNU, but it is also, arguably, a valid choice, since GNU is more analogous to the UNIX System than Linux is.
In contrast to Linux, the open-source BSD derivatives continue the UNIX duality of being both full systems and operating systems. They have borrowed some important tools from GNU (replacing various BSD UNIX tools that were removed for copyright reasons), but have always been distinct systems from GNU itself.
Sun used to call Solaris an 'operating environment', and SunOS the underlying 'operating system', but this is no longer true. See: Solaris Operating System
'Operating system' has long been used in the broad sense of an entire system, like UNIX (kernel and userland) or Microsoft Windows, so Sun presumably decided to stop swimming against the tide over a pedantic issue that most people don't care about.
Red Hat Linux is conceptually similar to Sun Solaris or Microsoft Windows. Linux itself would be better compared with the SunOS kernel (part of Sun Solaris) or the NT kernel (part of Microsoft Windows).
Arguing about whether an OS is a narrow (referring to, for example, Linux, SunOS and NT) or broad (referring to, for example, Red Hat Linux, Sun Solaris and Microsoft Windows) is just a game of semantics, and beside the point: Schwartz was referring to three comparable systems, and was broadly correct, although I think SuSE LINUX (I don't know why they insist on spelling it in all capitals) is still in the game.
If Linux vendors like Red Hat apply their own changes to the kernels they distribute (differentiating them from the baseline Linux releases), each is clearly a different system from the others (like different versions of UNIX, but with the changes shared). Even if the kernels are identical, the variations in software layered above arguably differentiate the systems enough to view them as separate. Either way, from a user perspective, Red Hat Linux and Gentoo Linux, for example, seem less alike in many respects than FreeBSD and OpenBSD (which are unarguably different systems, but both based on 4.4BSD, with a lot of code sharing).
Slashdot Mirror
It actually does make sense if the delay in completing the MSH environment is down to providing command line interfaces to every aspect of the OS, as opposed to completing the shell itself. MS Exchange has a much smaller set of features to provide command-line interfaces to than the entire MS Windows system.
You can get that today, in Window XP 64-bit Edition for x64 (or some name like that), which is based on the NT 5.2 kernel, like Windows Server 2003. The only problem with it is device drivers: if all your hardware is supported by drivers from Microsoft, it's fine. If not, you'll have to wait for the hardware vendors to port the required drivers to 64-bit (some have already, whilst others won't commit to ever doing it).
I think AppleScript is more like the existing Windows Scripting technologies. The difference with MSH is that it's an attempt to merge that sort of thing into an interactive command-line environment, and include every aspect of the system that the GUI can manage. (I don't know if AppleScript covers every aspect of the system that is accessible from the GUI, but I know Windows Scripting doesn't.)
Unix administrators like managing their systems from a command-line rather than graphical interface, and a lot of them are moving from Unix/Risc to Linux/x86/amd64, instead of to Windows/x86/amd64, as Microsoft would like.
In other words, this new shell isn't so much for current users, and certainly not for desktop users, as it is to attract Unix and Linux administrators to Windows. Microsoft Services for Unix helps, by offering a Unix command line and allowing Unix software to run on the NT kernel, but the problem is that Windows services can't be managed from the Unix command line.
In general, I like the 'keep it simple' philosophy of traditional UNIX, but I'll try MSH, when it gets to a more mature stage, before deciding whether or not Microsoft have come up with a better CLI than the simple UNIX model.
Win32 was actually released with Windows NT 3.1, in 1993, which makes it 12 years old. NT development started in 1988 or so, but I don't think the Win32 subsystem was started until about 1990, round about the time Microsoft and IBM parted ways (because Microsoft wanted Win32, instead of OS/2, to be the primary API/subsystem for NT).
The conflicting claims make EDS look a bit stupid, but I don't think most objective people think criticism of Linux 'shows religious bias', even if it's vague and perhaps invalid.
'Port' means the version of the software for a given platform. Opera is very portable (or 'cross-OS capable' as you say), which is why there are so many ports!
Nokia probably don't want to put all of their eggs into a single basket, as the saying goes. If the difference between the value added by Opera and the price of Opera is greater than the value added by Minimo, minus any funding from Nokia, then Opera is a better choice. If not, Minimo may be the better one.
Opera runs on most PC and PDA/mobile OSes, including Windows Mobile. Nokia, however, don't offer any phones running Windows Mobile, so you're right that the Nokia phones the article refers to run Symbian OS.
The article doesn't make clear why there will be up to a month delay. It may mean Microsoft will simply test on the configurations used by American government customers first, which wouldn't really delay patches for the rest of us, so wouldn't be so bad.
If this policy is only for the American government, that bothers me more, but not enough to use an inferior substitute. If it's for all governments, and doesn't delay getting patches to other users, I don't care at all.
Stallman is also very political, and alienates a lot of people (like me) with his divisive rhetoric. I've nothing against Microsoft, Apple, Sun or anyone else who would like to sell proprietary software. I've nothing against those who would like to give away software either, and would rather have source code, but only if those who wrote it want to give it to me. I therefore find Stallman's 'software war' mentality alien and off-putting.
Linux is just a kernel, and would not exist without GNU, but without Linux, FreeBSD would probably be the dominant open source OS today. Stallman should be grateful that Torvalds promoted his project via the Linux kernel, because it would otherwise have never caught on, and the GPL would probably be far less popular than it is.
Upgrading a Dell or HP to XP Professional would probably still leave it cheaper than a similar Mac. Some of us also prefer the Windows XP GUI to Mac OS X one, and appreciate all the applications available for Windows XP.
My favourite window manager is ion, but the rest of the GUI on Unix-like systems is pretty random, so Windows XP is best overall for me (although I would really like something like ion on it). I think the Mac OS X command line is a bit better than Interix on XP (mostly because more tools have been ported to it), but either one will work for me.
With lots of pointless eye candy and all the spyware and viruses you can eat. :P
I've used Windows XP since it came out, and I've never had any virus/worm/spyware infections.
Certification is simply a guarantee by IBM and Red Hat that a given application meets a set of requirements which they (IBM an RH) have defined. This allows potential customers to look at a set of certification criteria and, if the criteria meets their needs, trust that the associated list of certified applications will do the same.
Partnering with multiple Linux vendors will help prevent any one becoming dominant in the market. This, in turn, will prevent the Linux vendors adding too much value to the operating system, which, in IBM's view, should be a commodity layer for running IBM server applications, supported by IBM services.
I don't know if it's the same thing, but the link in the parent post says this:
HP's partership with Intel to produce the Itanium was an attempt to produce a replacement for PA-RISC (which, with Alpha, was always a top performer in the RISC market) with wider market appeal. However, we all know how that's been going...
Another interesting thing is that open source doesn't eliminate the 'network effect', as it's called in economics, which has been central to Microsoft's dominance of the operating systems market. The more people there are using Red Hat Linux, the more valuable it becomes, and the availability of source code doesn't change that. It makes it easier for competitors to interoperate with Red Hat Linux, but if Red Hat Linux had sufficient market share (e.g. similar to Microsoft Windows), it's quite possible that competitors would find it extremely difficult to get the market to accept alternatives to it.
As anyone familiar with the rationale behind organisations like ISO and The Open Group will know, open source and open standards are two different things. Both are important, but, at the end of the day, I think open standards are very much more so.
Windows CE is a very different thing, but it's a limited system designed for small devices, and so isn't relevant to a discussion of full desktop/server systems like Microsoft Windows (NT), Sun Solaris or Red Hat Linux.
Linux is comparable to the UNIX Operating System, not the UNIX System, but, following UNIX tradition, most people refer to both the OS proper and the entire system as simply 'Linux', even if, techically, only the former is. The ugliness of the GNU name (both to the eyes and the ears) may be part of the reason most people don't call it GNU, but it is also, arguably, a valid choice, since GNU is more analogous to the UNIX System than Linux is.
In contrast to Linux, the open-source BSD derivatives continue the UNIX duality of being both full systems and operating systems. They have borrowed some important tools from GNU (replacing various BSD UNIX tools that were removed for copyright reasons), but have always been distinct systems from GNU itself.
'Operating system' has long been used in the broad sense of an entire system, like UNIX (kernel and userland) or Microsoft Windows, so Sun presumably decided to stop swimming against the tide over a pedantic issue that most people don't care about.
Red Hat Linux is conceptually similar to Sun Solaris or Microsoft Windows. Linux itself would be better compared with the SunOS kernel (part of Sun Solaris) or the NT kernel (part of Microsoft Windows).
Arguing about whether an OS is a narrow (referring to, for example, Linux, SunOS and NT) or broad (referring to, for example, Red Hat Linux, Sun Solaris and Microsoft Windows) is just a game of semantics, and beside the point: Schwartz was referring to three comparable systems, and was broadly correct, although I think SuSE LINUX (I don't know why they insist on spelling it in all capitals) is still in the game.
If Linux vendors like Red Hat apply their own changes to the kernels they distribute (differentiating them from the baseline Linux releases), each is clearly a different system from the others (like different versions of UNIX, but with the changes shared). Even if the kernels are identical, the variations in software layered above arguably differentiate the systems enough to view them as separate. Either way, from a user perspective, Red Hat Linux and Gentoo Linux, for example, seem less alike in many respects than FreeBSD and OpenBSD (which are unarguably different systems, but both based on 4.4BSD, with a lot of code sharing).