Subject: What's LINUX? And other misc questions....
Date: 4 Sep 92 06:49:58 GMT
First off, I looked for a FAQ list. The only thing I found was information to
sites that had the LINUX source. This may be useful down the line, but not
right now. Other searches such as a thread on "Linux history," more
specifically, followups to the latter, seemed like a personal conversation
and yielded nothing.
Specifically, I would like to know what LINUX is? Yes, I know an operating
system. Yes I know, very similar to UNIX. But what's the catch or impetus for
people to own it? How old is it? How stable?
There's MINIX which offers you all the source to the OS and utilities. There's
Coherent, now in its fourth version, supporting COFF executables and a 32
OS and compiler. Not to mention all its utilities and other facilities, such
So, iterating, what's LINUX's tune? Why should one bother with it now that
you can get BSD's UNIX for the 386 for free? MINIX and Coherent looked good,
however personally I don't see much motivation to use them now that you can
get the "Real McCoy," 386BSD UNIX for free.
So what does LINUX offer that these other operating systems don't?
Please respond directly, I would appreciate it. Thanks.
From: firstname.lastname@example.org (Drew Eckhardt)
Subject: Re: What's LINUX? And other misc questions....
Date: Fri, 4 Sep 1992 17:20:47 GMT
In article <192248.014958U20565@uicvm.uic.edu> writes:
>First off, I looked for a FAQ list. The only thing I found was information to
>sites that had the LINUX source. This may be useful down the line, but not
>right now. Other searches such as a thread on "Linux history," more
>specifically, followups to the latter, seemed like a personal conversation
>and yielded nothing.
>Specifically, I would like to know what LINUX is? Yes, I know an operating
>system. Yes I know, very similar to UNIX. But what's the catch or impetus for
>people to own it? How old is it? How stable?
Linux is free. Linux is small. Linux has been fairly stable for quite some
time, with new releases being almost entirely new features (Such as a modern
VM, with unified buffer cache and user memory pool, extended / DOS
file systems, etc). I first used Linux last November/Decemeber, and since then
it's become quite stable.
>There's MINIX which offers you all the source to the OS and utilities. There's
Minix also costs money. As it comes out of the box it can't run
much of the freely available software, because of process size
restrictions and the lack of VM.
There are patches available that rectify this situation, but it won't
be as clean, or easy, as starting with a real 32 bit VM OS from the
start, like Linux.
>Coherent, now in its fourth version, supporting COFF executables and a 32
>OS and compiler. Not to mention all its utilities and other facilities, such
Coherent also sells for $99, and there's no way you'll get source code for it.
While it is a 32 bit OS, it doesn't have VM.
>So, iterating, what's LINUX's tune? Why should one bother with it now that
>you can get BSD's UNIX for the 386 for free? MINIX and Coherent looked good,
But can you use the Jolitz port of BSD?
Linux is significantly more stable the the Jolitz port of BSD. Spend some
time over in the comp.unix.bsd group, and you'll find a lengthy list of known
bugs that MAY be fixed in the next release.
BSD runs on less hardware than Linux - , Linux
runs on IDE, RLL, MFM, ESDI, Adaptec 154x, 174x, Seagate, Ultrastor 14f, Future
Domain 16 bit, and Western Digital 7000FASST SCSI disks, does enough 387
emulation on hardware lacking a FPU to use a GCC with hardware floating
point support, and has extended text mode support in the console driver
for most SVGA boards plus CGA, MDA, Herc, EGA, and VGA.
BSD also requires more resources than Linux - without Linux's
shared libraries, X applications can grow from < 100K to > 400.
Bill is also running a one man show - which makes it very difficult to get
neat features into the kernel. On the otherhand, Linus will integrate
any netter-contributions provided that there are clean context diffs
and they fit cleanly into the Linux scheme of things. Consequently,
Linux has lots of nifty features that BSD lacks, such as shared libraries,
the MSDOS file system, virtual consoles that emulate VTxxx's, etc.
>however personally I don't see much motivation to use them now that you can
>get the "Real McCoy," 386BSD UNIX for free.
>So what does LINUX offer that these other operating systems don't?
It's smaller, has more features, runs on more hardware, consumes less
resources, and is more stable.
Microsoft is responsible for propogating the evils it calls DOS and Windows,
IBM for AIX (appropriately called Aches by those having to administer it), but
marketing's sins don't come close to those of legal departments.
Boycott AT&T for their absurd anti-BSDI lawsuit.
From: email@example.com (William F. Jolitz)
Subject: Re: What's LINUX? And other misc questions....
Date: 4 Sep 1992 19:43:20 GMT
In article <1992Sep4.firstname.lastname@example.org>
email@example.com (Drew Eckhardt) writes:
>But can you use the Jolitz port of BSD?
Well, we've got over 250,000 copies out there, according to UNIX Magazin.
There is a 17 part feature series documenting many elements of the internals
in both English and German. We have automated installation procedures
which allow one to qualify, partition, download, and install/extract the
binary, source, and additional software packages. We've had tremendous
success with this approach, which accounts for it's presence throughout
the world in a matter of a few months entirely through the network.
A number of key research projects are using elements of 386BSD in their
own work. And a book on the internals will be out soon.
By the way, the "Tiny 386BSD" installation floppy is also part of a
charity program run under the auspices of the editors of Dr. Dobbs Journal.
In a nutshell, when you send in your floppy and SASE mailer to get a
free copy of Tiny 386BSD, you can send in a dollar or two to go to the
Children's Support League, which funds small charities which help
disabled children talk via computers, abused children get respite weekends,
disadvantaged children find a safe place to play, and so forth. There
is no obligation, but already hundreds of readers have sent in cash,
checks, and money orders throughout the world -- some with notes saying
"for the kids". So a small charity is also being helped by the generosity
of the user community.
I'd say that there is certainly a great deal of use coming from 386BSD,
wouldn't you? And a bit of good works as well.
>Linux is significantly more stable the the Jolitz port of BSD. Spend some
>time over in the comp.unix.bsd group, and you'll find a lengthy list of known
>bugs that MAY be fixed in the next release.
Not necessarily. I've been running 386BSD for many months without serious
incident. Of course, some people are stress-testing the system, and we
don't work perfectly on all configs, but neither does LINUX or MACH, for
that matter. I'd say we're all very much in a state of change and as each
system is enhanced and developed, they will get better. There is an
enormous amount of untapped talent out there, waiting for an opportunity
to try something new. In the 6 months that 386BSD versions have been
available, I have been impressed with the ability, creativity, and
enthusiasm of users, and I have no doubt that will continue.
>BSD runs on less hardware than Linux - , Linux
>runs on IDE, RLL, MFM, ESDI, Adaptec 154x, 174x, Seagate, Ultrastor 14f, Future
>Domain 16 bit, and Western Digital 7000FASST SCSI disks, does enough 387
>emulation on hardware lacking a FPU to use a GCC with hardware floating
>point support, and has extended text mode support in the console driver
>for most SVGA boards plus CGA, MDA, Herc, EGA, and VGA.
As per the 0.1 Release notes:
"This release supports a 386/486 SX/DX ISA (ATBUS) system, with the traditional
hard and floppy disk controller (MFM, ESDI, IDE, SCSI - Adaptec 1542, RLL).
Also the usual display adapters (MDA/CGA/VGA/HGC) are supported, along
with the serial communications ports (COM1 and COM2) and parallel port (LPT1).
Ethernet controllers supported are the Western Digital/SMC 8000/Elite series,
Novell NE1000/2000, 3COM 3C503, and the ISOLAN ISOLINK. Clones also appear
to work quite well. Tape drive support is available for QIC-02 and SCSI
(Adaptec 1542) controllers as well, allowing use of 3M cartridges of QIC-60
through QIC-150 format (no QIC-40 or QIC-80 supported at this time). SCSI
(Adaptec 1542) CD-ROM drives and CD-ROM ISO 9660 filesystem format is also
supported. The system supports either 287 or 387 coprocessor, but does not
require one for operation."
X386 (now known as Xfree86) supports several standard SVGA chipsets
and we have recently gotten it supported on standard hardware. Contact
firstname.lastname@example.org for information on the X386 project.
There is currently extensive work ongoing to support ISDN and X.25 cards,
high-speed networking applications using fibre-optic, as well as the
Microchannel architecture. In addition, there are software projects focussing
on LWP's and multiprocessing, performance evaluation, filesystem i/o
bandwidth improvement, virtual memory design, and a new object-oriented
approach to streams called CURRENTS. Several of these projects
are being coordinated at different universities and others are directly
coordinated by us. However, most of these improvements/projects are
only relevent to the research community.
>BSD also requires more resources than Linux - without Linux's
>shared libraries, X applications can grow from < 100K to > 400.
This is true -- we have only recently formed a research group on shared
libraries. Up until now, we have been able to focus on other more critical
areas of the system, but with the recent advent of X, it has become necessary
to add shared libraries.
As 386BSD is intended for research/educational purposes, and as a development
system, we have proceeded cautiously on shared libraries. Shared libraries
are easy to "hack" into the system, but once you do, you must carry any
legacies with you. That is not the way that BSD is done -- it gets incorporated
into too many serious efforts to be done cavalierly. Granted, it's a great
responsibility, but we take it seriously and also listen to serious suggestions
on this and other topics first.
>Bill is also running a one man show - which makes it very difficult to get
>neat features into the kernel.
Well, let's think about this. We provide source, and a directory
for contributions, tests, and new ports of software. It is very active.
People seem to recompile and post changes to both the mirror sites and
the newsgroups regularly, so they don't seem to be limited in this
And in the 0.1 Release Contributor List, we cited 39 key people who
did major fixes/contributions for 0.1 (including, by the way, Linus
Torvalds for the use of his emulator -- he was most gracious in this
regard, and as we modify it we send changes back his way for his use
if he so desires). This list does not include a lot of minor fixes and
approaches. They all did very good work.
The next patch of the system will have a number of new names and lots
of new features along with the usual "already known" fixes.
Large projects such as TEX, internationalization of the keyboard,
ISDN and X.25, high-speed networks, X386, and so forth are done
out of house with input from us on coordination.
Doesn't sound like a one-man show to me. A lot of people have put a
great deal of time into this project, not to mention those who have
done work over the years building the wealth of BSD software (such
as Macklin's NFS) which has been made accessible. It's a great project,
with many man-years involved. As I stated earlier, it's a research
project -- not a commercial project or a hobby.
>On the otherhand, Linus will integrate
>any netter-contributions provided that there are clean context diffs
>and they fit cleanly into the Linux scheme of things.
As do we. Not weekly, however. Too frequent changes lead to great
problems within the BSD community. Perhaps this is not a problem with
>Linux has lots of nifty features that BSD lacks, such as shared libraries,
>the MSDOS file system, virtual consoles that emulate VTxxx's, etc.
And we think that's great! If it gets people to try new things, so
much the better. Instead of an either/or attitude, perhaps we should
try to learn from each other's experiences. (For example, for the
tcp/ip networking project for LINUX, one might try to leverage the
expertise and code evolved from the Berkeley networking paradigms).
If some people feel more comfortable with LINUX, the more power to them!
If other people want to do research and desire 386BSD, that's great too.
And if you want to use both, you can! 386BSD runs in a partition too!
Each system meets different needs and audiences. We can learn from each
other's approaches, successes and failures. After all, we're all here to
learn, aren't we? :-)