When is it worth saying it's Linux? (The H) [LWN.net]

When is it worth saying it's Linux? (The H)

Posted Jan 31, 2010 23:21 UTC (Sun) by Tara_Li (guest, #26706)
Parent article: When is it worth saying it's Linux? (The H)

A lot of this seems to come down to a very blurry line between "application" and "operating system". At one point, the OS was what directly affected the hardware - controlled access to memory, hardware, etc. Then came shells and what MS-DOS called "external commands". Things that were little more than wrappers around OS calls, presenting them in easy to use fashion. And these got glued on, and called part of the OS. Now, we bitch and complain about MS calling "Internet Explorer" part of the OS, while including three different browsers, three different office suites, and dozens of games, graphics utilities, and other applications as all part of the Linux OS.

I say we go back to the real thing. If it touches the hardware directly, it's the OS - the kernel plus drivers. Add anything else, and it's a distro - Windows XP distro coming in a half-dozen flavors, Windows Vista distro, Windows 7 distro, Fedora, Redhat, BSD distro, etc. bash is *NOT* a part of the OS - it's a command shell *APPLICATION*. gcc - a compiler *APPLICATION*. They're not part of the OS - they can all be replaced, and it's still the same OS. Android? It's an *application* on top of Linux, with other applications on top of it!

If we don't draw a line somewhere, eventually there will be no applications - just a few insanely bloated Operating Systems in huge numbers of permutations - but all legitimately calling themselves "operating systems"

to post comments

When is it worth saying it's Linux? (The H)

Posted Feb 1, 2010 1:32 UTC (Mon) by AndreE (guest, #60148) [Link] (1 responses)

Yes but a kernel != operating system.

Sure, the kernel manages resources and provides a nice abstraction and interface to those resources, but so does a "third party" GUI toolit, or external library, or shell. Just because something runs in user mode doesn't mean it isn't part of an operating system. I mean, any interactive system needs some tools that can actually make use of the kernel. Hence why we have the actual distinction between a kernel, which we can say provides access to and manages a machine's resources, and the actual software that allows us to "operate" and utilise the system. Making a blanket claim that binutils or coreutils or indeed bash are not part of the operating because they are merely wrappers to system calls or plain applications doesn't make much sense because linux as it's used in 99% of instances would not be very useful or operable without these out-of-kernel applications. Of course, for my mp3 player, it makes a lot of sense to suggest that the operating system is what is directly managing the hardware.

I also don't really understand the panic around this particular naming convention anyway. I hardly see kernel devs or gnu-chain devs adding bloat just because they are confused as to what the "official" definition of an operating system is.

When is it worth saying it's Linux? (The H)

Posted Feb 1, 2010 14:05 UTC (Mon) by sorpigal (subscriber, #36106) [Link]

All I can say is: "This debate again?"

Everyone knows what a kernel is but no one can decide what an operating system is. Some people like to say "kernel == OS" because it's so much simpler than trying to get agreement on anything else. Everyone can agree that the kernel is part of the OS, but beyond that you get a debate with lines drawn up every which way.

There will never be an end to it and there will never be a single accepted definition of OS. Can we agree on that and work on something constructive? We need a way to discuss these things--and refer to them in documentation and marketing material--without a lot of ambiguity. Can we agree that OS--however *you* want to define it--is a bad term for this?

That, at least, would be progress.