What is the basic/technical difference between windows, Linux, Mac?

[jpollard]

Originally Posted by Antrikssh

It's not a fight guys just chill and send your views so that our discussion will be helpfull to all those who suffer for this question like me. All the post I read but no one explane me answer of basic question like :-

1> Linux is all about file then, windows and mac work on what?

Linux and MacOS follow the UNIX style - everything is a file, device access, command access. The GUI is a wrapper for command access. Where such wrappers are NOT used then things get very cumbersome, like Gnome3 (though not limited to it). All command access is standard - a process receives the path to program and the name it was invoked by with a list of parameters. Shells are (normally) a simple programming language with a simple command invocation (pathname, parameters) passed to the program. It is possible to make complex command interpreters.. but not usually worth the effort. GUI interpreters are the most complex of them.

Windows uses a command interpreter to translate commands/menus into actions to start programs with specific parameters. One program may be invoked for multiple commands, just with different option parameters.

2> Kernel of all the o/s build by using c, c++ then what is the basic things that make all o/s different from each other?

Efficiency, Both Linux and MacOS are based on the UNIX design, where MacOS is close to the BSD design on top of another core. Linux has a greater freedom to experiment with new features/capabilities where both Windows and MacOS are limited to what the vendors decide. Linux and (I believe) MacOS uses similar interprocess communication. Added IPC methods (such as dbus) make things more complex, but address certain generality capabilities, but applications are not required to use them.

Windows makes the complex IPC mandatory for IPC.

3>why Linux is most preferable o/s for server?

Efficiency, ease of configuration, ease of debugging configurations (well IMO up to F15/16), replication of configurations. And low license fees. It also has the freedom to configure it in any way desired. Both Windows and MacOS can only be configured the way the vendors decide is appropriate.

4>Linux used 319 system calls then how many in windows and mac?

Not really useful - there are many MANY more system calls possible in any UNIX like system - see ioctl for an example - each is different, so there are multiple system calls being windowed through one invocation.

BTW, according to /usr/include/bits/syscall.h, there are only 28 system calls for users. Others may be hidden (ioctl type thing), but many are being dropped by the creation of special purpose filesystems (such as /proc, /sys, devfs...) that bypass the need for lots of system calls. These use the existing system calls (open/close/read/write) to implement many operations that used to require system calls.

To my knowledge, neither MacOS nor Windows support doing things this way, though MacOS does have the proc filesystem.

And many more question are strick in my mind when I read about any o/s(Linux, Windows, Mac). All the post that you send most of them are easily find out on net but this type of information that i mention is not found on net until you read all this o/s on depth.

That is mostly because it isn't that important.

Originally Posted by Antrikssh

It's not a fight guys just chill and send your views so that our discussion will be helpfull to all those who suffer for this question like me. All the post I read but no one explane me answer of basic question like :-

1> Linux is all about file then, windows and mac work on what?
2> Kernel of all the o/s build by using c, c++ then what is the basic things that make all o/s different from each other?
3>why Linux is most preferable o/s for server?
4>Linux used 319 system calls then how many in windows and mac?
And many more question are strick in my mind when I read about any o/s(Linux, Windows, Mac). All the post that you send most of them are easily find out on net but this type of information that i mention is not found on net until you read all this o/s on depth.

Antrikssh,

You have asked a very nice, but, a very open ended question. But, your reply that I have quoted above gives some hints on the type of answer you were expecting.

Superficially, you have found all the answers that you asked for:
* Linux is open source, Windows (the core kernel and a lot of system apps) are NOT open source
* Mac OS X is essentially Mach-with-FreeBSD/XNU hybrid kernel
* Windows ABI has been stable for a long time (I didn't know that)

I am going to make an attempt at answering your question with the hope that it satisfies you, but, I make no guarantees about it.

So, any OS is made up of choices. The choices define various things:
* size of the kernel
* contents of the kernel (minimal, just-enough, more than enough, much more than enough, etc)
* functionality provided by the kernel
* the algorithms chosen by the various kernel subsystems (scheduler, memory manager, process model, threading model, file system abstraction, actual file systems supported, etc)
* the choice on how devices are accessed/accessible from the user-land
* the networking stacks supported by the system
* the architecture of the networking stacks (e.g. Unix SVR4 used the STREAMS architecture to build networking stacks, while the TCP/IP stack was an integral part of the *BSD systems, Linux uses its own software architecture for the networking stack and I am sure Windows and Mac OS X have their own)
* Etc
* Etc

Clearly, to compare the various OS-es (even just the three that you have given) would make a good grad level project for some one, but, giving out all the differences would be practically tough on a forum.

Finally, I don't think the "number of system calls" and the implementation language, etc of the kernel matter (or should matter). Practically 90% if not more hard-core systems programming still happens in C and will happen in C for a foreseeable future. As for system calls, you and I could add as many system calls as we want. I guess, what the system call actually ACHIEVES is more important than how many there are. But, that is getting into philosophy and you don't want a philosophy lecture do you?

As for why is Linux the preferred "server" OS? I feel the following reasons should answer that question:
* Linux is Open Source. So, if you find a bunch of good Linux hackers, you can "transform" it into anything you want.
* Linux is FREE. Anyone and everyone can download it, hack it, and use it as they see fit. You don't have to pay licensing fees to any company or organization. So, a company that runs a large social networking site, or a company that runs an entire online content-farm (think news papers, high-end blogs, etc) would clearly want Linux on their servers. Because you may start with 2 servers, but, if the world loves your content, they are going to want more of it, and you are going to want more servers from which you serve your "much loved" content. And you definitely DON'T WANT TO PAY licencing fees as you add more and more servers.
* Linux is portable. Linux already runs on x86 32 bit, x86 64-bit, ARM, POWER, MIPS, and other "contemporary" processors. Linux has always supported other weirdo processor architectures like Alpha, Itanium, etc. And with all the brainiacs who hack the Linux kernel, we all (and all large corporations) can be sure that Linux will work on any new processor that comes along for a foreseeable future.
* Most of the popular server apps (web servers, mail servers, DNS servers, database servers, caching servers, etc) already work "optimally" on Linux. (The word "optimally" is debatable, but, let's go with the flow for now).
* Linux is ductile and malleable. You come up with some new storage scheme or a new networking protocol or a networking stack, you can write code for your shiny-new-subsystem and release it out into the wild for people to play with. So, if you come up with a new server architecture, you can be pretty sure you can hack the Linux kernel to fit your server. Again, the combination of "OPEN SOURCE" with "ABSENCE of licencing fees" makes Linux a great solution.

Of course, most of those postives are applicable to systems like *BSD (e.g. my favorite, FreeBSD), but, my guess is that Linux has a much larger "work-force" than those other systems. In fact, if I understand it correctly, till just a few years back Yahoo, Hotmail and many other web based services used FreeBSD as their preferred server OS. I can also assure you that a few "high-end" router/switch vendors have-used/are still using FreeBSD as the core OS on their networking systems.

Anyway, I hope my brief answer satisfies some of your queries. Finally, I don't understand what you mean by "linux is all about file".

HTH
Mahesh R S

[rclark]

I don't understand what you mean by "linux is all about file".

I think he referring to how most resources are treated as a 'file'. For example when you open a serial port, you 'open' a file ( such as /dev/ttyS0 ) which is streamed as a file (write/read).... All part of the '/' ' directory tree. When you want to get cpu info you 'cat /proc/cpuinfo' ... etc.

I think one major overall difference is that from the ground up Linux/Unix were designed as multi-user systems. Unlike Windows/Dos which has been a single user system from the ground up. We are not talking just like a file server where multiple people can be served files and access printers. We are talking actually logging into the machine (say a telnet/ssh session or via VNC) and using all the machines resources such as compilers, office apps, play games, etc, just like you were 'local'. The external computer becomes a 'dumb' terminal so to speak. My Windows XP box for example in my home network, doesn't have access to the internet (by design). A family member has to log into a VNC session (connect to one of the Linux boxes) to check their e-mail and browse the internet. We all can be on at the same time.

You could write a book on the standard posix API that Linux uses, and the MS API for writing applications for Windows. What is nice about Posix standards is I can recompile my application on any Linux box (or probably BSD as well) and it will do so, no matter what the flavor of Linux I am running.

A book could be written, but to me these are the major differences (besides open vs. closed, free vs. pay).

[Dan]

It's not a fight guys just chill and send your views so that our discussion will be helpfull to all those who suffer for this question like me. All the post I read but no one explane me answer of basic question like :-

Hmmm.

I think we've got a whole lot of good helpful people here who have been trying very hard to answer some basic questions that would probably be best served by some serious independent research, investigation and plain old homework by the original poster.

This one is very close to falling under the "homework" rule, and as such, being closed. Methinks the overall bottom line is, a lot less attitude is in order here.

Countdown started ...

Originally Posted by rclark

I think he referring to how most resources are treated as a 'file'. For example when you open a serial port, you 'open' a file ( such as /dev/ttyS0 ) which is streamed as a file (write/read).... All part of the '/' ' directory tree. When you want to get cpu info you 'cat /proc/cpuinfo' ... etc.

I think one major overall difference is that from the ground up Linux/Unix were designed as multi-user systems. Unlike Windows/Dos which has been a single user system from the ground up. We are not talking just like a file server where multiple people can be served files and access printers. We are talking actually logging into the machine (say a telnet/ssh session or via VNC) and using all the machines resources such as compilers, office apps, play games, etc, just like you were 'local'. The external computer becomes a 'dumb' terminal so to speak. My Windows XP box for example in my home network, doesn't have access to the internet (by design). A family member has to log into a VNC session (connect to one of the Linux boxes) to check their e-mail and browse the internet. We all can be on at the same time.

You could write a book on the standard posix API that Linux uses, and the MS API for writing applications for Windows. What is nice about Posix standards is I can recompile my application on any Linux box (or probably BSD as well) and it will do so, no matter what the flavor of Linux I am running.

A book could be written, but to me these are the major differences (besides open vs. closed, free vs. pay).

RClark,

Don't quote me on this, because I have never used "Windows" expertly, but, Windows at least since I knew it well: XP onwards, has always supported multiple users, no? Sure, you can't have multiple users logging into the system simultaneously, but, I think you can have multiple users on any given Windows system. But, again, don't quote me on that.

---------- Post added at 03:40 PM ---------- Previous post was at 03:38 PM ----------

Originally Posted by Dan

Hmmm.

I think we've got a whole lot of good helpful people here who have been trying very hard to answer some basic questions that would probably be best served by some serious independent research, investigation and plain old homework by the original poster.

This one is very close to falling under the "homework" rule, and as such, being closed. Methinks the overall bottom line is, a lot less attitude is in order here.

Countdown started ...

Dan, you spoke what was on my mind. Even I assumed that this was part of homework/assignment/thesis. That is why I clearly hinted at it in my first reply on this thread. But, being a new member, I did not want to "antagonize" other members of this forum.

Thanks for being a good administrator Dan.

[deanej]

Originally Posted by maheshrsblr

RClark,

Don't quote me on this, because I have never used "Windows" expertly, but, Windows at least since I knew it well: XP onwards, has always supported multiple users, no? Sure, you can't have multiple users logging into the system simultaneously, but, I think you can have multiple users on any given Windows system. But, again, don't quote me on that.

It does, but one could say that a system where only one user can be logged in at a time is just a slightly more fancy single user system with multiple accounts.

[Doug G]

Originally Posted by deanej

It does, but one could say that a system where only one user can be logged in at a time is just a slightly more fancy single user system with multiple accounts.

Nope. Windows NT was designed from the ground up as a multiuser OS. Every version based on the NT branch of windows is a multiuser OS (NT, W2000, XP, W2003, Vista, W7, W2008, etc.).

[jpollard]

And yet, you can't connect two users simultaneously.

Granted, the NT 3.5 release had more multi-user capability than is currently provided. but even then, it was serial multi-user. Only one at a time.

[deanej]

Originally Posted by Doug G

Nope. Windows NT was designed from the ground up as a multiuser OS. Every version based on the NT branch of windows is a multiuser OS (NT, W2000, XP, W2003, Vista, W7, W2008, etc.).

You still can have only one user doing work at a time. "Multi-user" in the Windows world just means that the OS can "shift gears" from one user's profile/permissions to another's. It doesn't allow multiple users to interact with the computer at the same time... theoretically this could be a criteria for a true multi-user operating system.

[Doug G]

Originally Posted by jpollard

And yet, you can't connect two users simultaneously.

Granted, the NT 3.5 release had more multi-user capability than is currently provided. but even then, it was serial multi-user. Only one at a time.

Sure you can. A 2nd administrative user connection is available via remote desktop on most workstation versions of windows, and terminal server on a server version of windows provides as many simultaneous users as your resources will allow.

[jpollard]

Originally Posted by Doug G

Sure you can. A 2nd administrative user connection is available via remote desktop on most workstation versions of windows, and terminal server on a server version of windows provides as many simultaneous users as your resources will allow.

So it takes additional software before it becomes multi-user...