Philip Rushik via plug on 21 Sep 2019 04:49:00 -0700

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] Richard M. Stallman resigns

>> On September 20, 2019 7:11:57 PM EDT, Steve Litt via plug
>> <> wrote:
>> >I see nothing ridiculous about it. It's accurate. If you want to use
>> >the word "Linux" for short, fine, but Linux is the kernel, and the
>> >other stuff is supplied by many others, with a good chunk of it being
>> >supplied by the GNU project.
>> Except that before rms, kernel and operating system meant the same
>> thing. We only make the distinction now because rms said they were
>> different.
> Where'd you get that information? I can't disprove it, but I doubt its
> veracity. See and note that
> MSDOS.SYS was DOS' kernel, first appearing in MSDOS 1.1. Look at
> to see
> that MSDOS 1.1 came out in May 1982. At
> you can see that the GNU project was
> first publicly announced in September 1983, more than a year after DOS
> 1.1. I can't find references, but I remember hearing about the "VAX
> kernel" in the 1980's. The word "kernel" wouldn't exist if it were a 1
> to 1 venn diagram match with "OS".

MSDOS is really a difficult case. Before DOS, an operating system and
a kernel by any definition _were_ the same, since operating systems
weren't packaged with applications, even the idea of hierarchical file
systems were in their infancy. DOS muddied the waters a little bit,
especially since all programs ran in the same address space as the
kernel, and the BIOS was
still available, so one could write DOS applications, or bare-metal
applications that didn't use DOS at all, yet ran under DOS, or even
applications that implemented their own hardware drivers.

Systems like MINIX can complicate things further, where core OS tasks
are not handled by the kernel, but by processes outside of the kernel
that the kernel communicates with. I can see an argument here for a
kernel being different than an OS.

Of course, Linux is much simpler and makes the division between OS and
application very clear, its a monolithic kernel, that runs in kernel
space, and applications run in user space.

> Looking at it another way, how many folks would call a kernel, with
> nothing else, as an OS?

This depends on what we were talking about, I imaging application
developers would. Certainly in your DOS example, an application
developer would only interact with MSDOS.SYS (unless he/she wasn't
very good at managing memory).

> Oh, and when I use BSD, I sure am glad they include ls, cat, grep, sed,
> AWK and the like.

Of course, those are all great. I also like using Geany, but that
doesn't make it any less of an application.

>> Also, not the gnu components are not that significant anyways, and
>> the don't really make up the operating environment like the kernel
>> does.
> Speak for yourself. Not a day goes by when I don't use cat, ls, AWK,
> and less multiple times.  I write whole programs with them. I use bash
> every day, and every shellscript I write starts with #!/bin/sh.

By this I did _NOT_ mean that the GNU components are not
useful/good/important. They just simply aren't enough of a core part
of the OS. Linux is responsible for loading, executing, and
multitasking those applications, GNU is responsible for tools that
make the system easier to use. That's nice, but its not a core part of
the system.

Think of it this way, I have a great idea for functionality X. If I
implement it under Linux, using c or bash or anything (doesn't
matter), I have created an application. If I want to implement it
without calling into Linux, my only option is to write a Linux kernel
module. In the latter case, my implementation of functionality X
becomes part of Linux, in the first case, its just an application, no
body would claim it becomes part of GNU.

> GNU can and has been written in musl and various library collections.
> Get Void Linux, musl style, and you'll have all those utilities created
> with musl.

This is not accurate, which is good for you because it undermines your
own argument. musl and busybox are not an implementation of GNU, they
are an implementation of POSIX standard utilities and libraries, just
like GNU is. If you assert they are GNU, then GNU is not GNU, its
POSIX, and there is no GNU OS. Perhaps we should call the OS
POSIX/Linux? Of course, this wouldn't make sense, because FreeDOS is
an implementation of MS DOS, yet we would never refer to it by a
concatenation of the two.

At the very least, I hope you wouldn't use GNU/Linux to refer to the
Void Linux distribution since it doesn't make use of GNU. Yet, is it
really a different OS?

> I don't find it all that painful to acknowledge the value of the GNU
> scaffolding in conjunction with the Linux kernel, nor to occasionally
> call it "GNU/Linux".

"Acknowledging the value" is not the same as what actually makes up an
operating system. GNU operates within Linux, other applications under
Linux have no choice but to interact with Linux, but making calls to
other utilities such as GNU ones are easily avoidable. GNU isn't
essential or core to anything, its just a collection of useful tools
and libraries.

GNU tools are historically important, that's undeniable, GNU
implemented tools that everybody wanted when there no other open
source options. It made Linux attractive to people that already knew
the tools. Just like Pro Tools makes OSX attractive to certain people.
But its really a stretch to call that the "Operating System".
Philadelphia Linux Users Group         --
Announcements -
General Discussion  --