Newsgroups: comp.os.coherent
Path: sparky!uunet!cs.utexas.edu!utgpu!utzoo!censor!meadow!alex
From: a...@meadow.uucp (Alex Dumitru)
Subject: linux
Message-ID: <1992Feb25.164208.13973@meadow.uucp>
Reply-To: a...@meadow.UUCP (Alex Dumitru)
Organization: Amdahl Canada Ltd., Software Development Center
References: <1992Feb21.213429.12224@ims.com>
Date: Tue, 25 Feb 92 16:42:08 GMT
Hello! There is frequent mention of 'linux' on this newsgroup as an
alternative to Coherent. Is this available by anonymous ftp?
From what IP address?
I have a 386 box that is lying around and it sounds like 'linux' is the
perfect OS for it. (Coherent seems happy on my 286)
If anyone is running this, I'd appreciate some feedback/comments!
Thx
Alex
--
Alex Dumitru - a...@bedrock.guild.org a...@biovision.utoronto.ca
a...@meadow.UUCP az...@amail.amdahl.com
Path: sparky!uunet!usc!wupost!darwin.sura.net!Sirius.dfn.de!fauern!unido!
cat!incom!orfeo!jrix!joachim
From: joac...@jr.sub.org (Joachim Riedel)
Newsgroups: comp.os.coherent
Subject: Re: linux
Message-ID: <1992Feb27.203717.15948@jr.sub.org>
Date: 27 Feb 92 20:37:17 GMT
References: <1992Feb25.164208.13973@meadow.uucp>
Organization: The European Coherent Support BBS
Lines: 16
a...@meadow.uucp (Alex Dumitru) writes:
> If anyone is running this, I'd appreciate some feedback/comments!
Watch the following newsgroup: alt.os.linux
Cheers,
Joachim
+----------------------------+--------------------+-------------------------+
| Joachim Riedel | joac...@jr.sub.org | j...@connie.de.convex.com |
| Geschwister-Scholl-Str. 48 +----------------------------------------------|
| D-6050 Offenbach am Main | European Coherent Support BBS: +49 69 858711 |
| Tel. +49 69 85 62 25 | V22bis, V32, V32bis (T3000 and USR Dual Std. |
+---------------------------------------------------------------------------+
Newsgroups: comp.os.coherent
Subject: Re: linux
Path: sparky!uunet!think.com!unixland!rmkhome!rmk
From: r...@rmkhome.UUCP (Rick Kelly)
Organization: The Man With Ten Cats
Date: Fri, 28 Feb 1992 13:44:15 GMT
Reply-To: r...@rmkhome.UUCP (Rick Kelly)
Message-ID: <9202280843.53@rmkhome.UUCP>
References: <1992Feb25.164208.13973@meadow.uucp> <1992Feb27.203717.15948@jr.sub.org>
In article <1992Feb27.203717.15...@jr.sub.org> joac...@jr.sub.org (Joachim Riedel)
writes:
>a...@meadow.uucp (Alex Dumitru) writes:
>> If anyone is running this, I'd appreciate some feedback/comments!
>Watch the following newsgroup: alt.os.linux
I subscribed to the Linux mailing list for awhile, and I am now reading
alt.os.linux. It has some very interesting features, but it is still
fairly rough.
A friend of mine has just started the process of installing Linux on his
386/33. It could easily get an award for being the hardest to install
pc operating system.
The process so far:
1. FTP two files called bootimage and rootimage.
2. Write the two files onto separate disks using a MSDOS utility called
rawrite.
3. At this point you can now boot a floppy based Linux system. Boot the
boot disk in drive a:, and then put the root disk in drive a: when
prompted.
4. You can then repartition the hard disk for Linux and MSDOS - one
partition for MSDOS and three for Linux.
5. You can copy files over to the new partitions, but you are stuck with
floppy drive a: being the root until you patch the bootimage disk in
some badly documented way to use the hard disk for root.
6. You can then bring over more of the Linux binaries through a very
convoluted method of getting them from MSDOS to Linux.
Mcopy is supposed to be on the root image, according to the docs, but it
is not. This makes the procedure of getting files from MSDOS diskettes
very uncertain.
There are lots of GNU sources ported to Linux, but you'd better have a lot
of memory. Since BASH has to take the place of the bourne shell, cnews would
be an immense drag on the system resources. ( /bin/sh = 30k, /bin/bash = 150k )
The system comes up in single user mode as login, init, and getty still have to
be implemented. It really needs UUCP as getting sources to the system is an
extreme pain in the neck.
It would probably be interesting to fool around with if you had an extra
386 box lying around doing nothing.
--
Rick Kelly r...@rmkhome.UUCP unixland!rmkhome!rmk r...@frog.UUCP
Path: sparky!uunet!mcsun!news.funet.fi!hydra!klaava!torvalds
From: torva...@klaava.Helsinki.FI (Linus Benedict Torvalds)
Newsgroups: comp.os.coherent
Subject: Re: linux
Message-ID: <1992Feb29.104418.681@klaava.Helsinki.FI>
Date: 29 Feb 92 10:44:18 GMT
References: <1992Feb25.164208.13973@meadow.uucp> <1992Feb27.203717.15948@jr.sub.org>
<9202280843.53@rmkhome.UUCP>
Organization: University of Helsinki
Lines: 38
In article <9202280843...@rmkhome.UUCP> r...@rmkhome.UUCP (Rick Kelly) writes:
>
>I subscribed to the Linux mailing list for awhile, and I am now reading
>alt.os.linux. It has some very interesting features, but it is still
>fairly rough.
Yes: Linux isn't really even in the same niche as coherent: coherent
seems to be quite "ready-to-run" out of the package. "Rough" is an
understatement when it comes to linux: if you don't know unix and PC's
then problems are /very/ easy to find.
>A friend of mine has just started the process of installing Linux on his
>386/33. It could easily get an award for being the hardest to install
>pc operating system.
Thanks :). You obviously haven't installed minix-386, not /that/ is
something that isn't fun (patches, recompilations of the kernel etc, but
the end-result is worth it). Installing linux is fairly straigtforward,
really, but the lack of documents is a pain (at least minix-386 comes
with a good tutorial - linux lacks almost all documetation).
>There are lots of GNU sources ported to Linux, but you'd better have a lot
>of memory. Since BASH has to take the place of the bourne shell, cnews would
>be an immense drag on the system resources. ( /bin/sh = 30k, /bin/bash = 150k)
Not really necessarily: linux uses page sharing to keep memory
requirements down: this also speeds up loading of things like bash that
are really mostly in memory anyway. I'm writing this under linux, and
my debug-key tells me I'm sharing 81 pages right now (one page is 4kB).
That's with 4 logins running, and 2 bash binaries and kermit.
But generally, yes, linux wants more from the system than coherent does
(a 386 for starters :), but my guess is that so will coh-386. Also,
most people are willing to pay for the extra memory usage when they get
all the features of bash. 4M is about minimum to really /use/ linux, and
8M is comfortable (but with swapping enabled, you can make do with 2M).
Linus
Newsgroups: comp.os.coherent
Path: sparky!uunet!cs.utexas.edu!uwm.edu!csd4.csd.uwm.edu!markh
From: ma...@csd4.csd.uwm.edu (Mark William Hopkins)
Subject: Re: linux
Message-ID: <1992Feb29.144603.9630@uwm.edu>
Sender: n...@uwm.edu (USENET News System)
Organization: Computing Services Division, University of Wisconsin - Milwaukee
References: <1992Feb27.203717.15948@jr.sub.org> <9202280843.53@rmkhome.UUCP>
<1992Feb29.104418.681@klaava.Helsinki.FI>
Date: Sat, 29 Feb 1992 14:46:03 GMT
Lines: 43
In article <1992Feb29.104418....@klaava.Helsinki.FI> torva...@klaava.Helsinki.FI
(Linus Benedict Torvalds) writes:
>>There are lots of GNU sources ported to Linux, but you'd better have a lot
>>of memory. Since BASH has to take the place of the bourne shell, cnews would
>>be an immense drag on the system resources. ( /bin/sh = 30k, /bin/bash = 150k)
>
>Not really necessarily: linux uses page sharing to keep memory
>requirements down: this also speeds up loading of things like bash that
>are really mostly in memory anyway. I'm writing this under linux, and
>my debug-key tells me I'm sharing 81 pages right now (one page is 4kB).
>That's with 4 logins running, and 2 bash binaries and kermit.
What I fail to see is why would someone even go through the effort of writing
150k worth of code when they could have done the same thing in five times less
space! Could there possibly be some hidden joy in unnecessarily bashing one's
head against the wall that I'm not privvy on?
This problem, for which I've coined the term "Software Cancer" is pandemic
to all GNU software and too much other public domain software, as I've
discovered in actually untangling some of the software mess myself. You
can always tell what is afflicted by Software Cancer by looking at the source.
In it is the symptom known as Comment Pollution: the result of the programmer
writing excessive and unnecessary verbosity describing the code that would have
been written in the first place had he or she been straightforward.
I mean do people actually write things 5 to 10 times bigger than necessary
on purpose, for security? for secrecy? to make it impossible to use?? or what??
And as for Linux: I cannot see justifying the use of so much RAM as to make
running this system impossible on the vast majority of 386 systems without
the purchase of memory extensions. If I have to buy something, then I might
as well just go out and purchase a Coherent instead.
Please, if you have a priority list, put these three things on the top of the
list:
(1) rewrite the kernel so that RAM usage is cut under 500kB.
(2) contact someone who's known for developing high quality
software tools (e.g. the producer of the Power Mix C compiler)
and see if they won't participate in the major contribution to your
public domain, with free software.
(3) write some good documentation in the implementation of Linux, as
the written word of a designer's intent has value to a software
developer far in excess to any value the source itself may have.
Path: sparky!uunet!mcsun!news.funet.fi!hydra!klaava!torvalds
From: torva...@klaava.Helsinki.FI (Linus Benedict Torvalds)
Newsgroups: comp.os.coherent
Subject: Re: linux
Message-ID: <1992Feb29.202713.7994@klaava.Helsinki.FI>
Date: 29 Feb 92 20:27:13 GMT
References: <9202280843.53@rmkhome.UUCP> <1992Feb29.104418.681@klaava.Helsinki.FI>
<1992Feb29.144603.9630@uwm.edu>
Followup-To: alt.os.linux
Organization: University of Helsinki
Lines: 118
I changed the flollow-up: to alt.os.linux - I guess most coherent users
aren't that interested (and if they are, they can always try to read
a.o.l).
In article <1992Feb29.144603.9...@uwm.edu> ma...@csd4.csd.uwm.edu (Mark William Hopkins)
writes:
> > [ bash under linux ]
>
>What I fail to see is why would someone even go through the effort of writing
>150k worth of code when they could have done the same thing in five times less
>space! Could there possibly be some hidden joy in unnecessarily bashing one's
>head against the wall that I'm not privvy on?
You are exaggerating a bit here: yes, bash is 5 times the size of the
bourne shell, but isn't exactly a re-implementation of it: bash does /a
lot/ of things the standard shell doesn't.
The joy of "bashing one's head against the wall" is in nice features
like aliases, functions, cursor keys, job control (yes, linux has it,
and bash supports it) etc - yes the binary gets a lot bigger, but it's a
real pleasure to use compared to the ordinary shell. And yes, linux
could have a simple /bin/sh and a more complex /usr/local/bin/bash, but
with page sharing it's usually a waste of memory to have two programs
that do the same thing.
[ I did have the minix /bin/sh running under linux way back in version
0.01 (last august or so), but I quickly changed. Hands up everyone who
doesn't like the lack of history in the bourne shell ]
>This problem, for which I've coined the term "Software Cancer" is pandemic
>to all GNU software and too much other public domain software, [ some deleted ]
GNU software is usually bigger than the "ordinary" one, and it does use
more memory. But most people who use GNU software think that it's worth
it: I know /I/ couldn't live without gcc (small exaggeration :). The
reason is, I think, that people who write lots of code often do it on a
big machine, so they have no reason to write the code to fit in a
smaller one. It shows up more clearly in freely distributable code:
proprietary programs have the constraint of trying to be sold, so they
want to fit on every machine.
> You
>can always tell what is afflicted by Software Cancer by looking at the source.
>In it is the symptom known as Comment Pollution: the result of the programmer
>writing excessive and unnecessary verbosity describing the code that would have
>been written in the first place had he or she been straightforward.
You'd like the linux kernel sources, I think: not many comments in
sight.. The kernel itself is pretty small (currently 180kB binary, and
that contains pageing, demand-loading, job control etc), but to get nice
performance, you need a lot of buffer pages etc, so linux actually wants
at least the lower 1 meg for kernel memory.
It's a case of "I want good performance, f**k the rest" :) I had minix
before I started on linux, and I got /thoroughly/ disgusted with
programs that fit in 64kB. I /want/ the added functionality bought at
the expence of a meg or two.
>And as for Linux: I cannot see justifying the use of so much RAM as to make
>running this system impossible on the vast majority of 386 systems without
>the purchase of memory extensions. If I have to buy something, then I might
>as well just go out and purchase a Coherent instead.
Well, I don't know about the rest of the world, but here in Finland at
least, 4M is about minimum configuration for a 386. And that's one of
the cheap taiwanese clones (I should know - that's what I bought, even
though I upgraded to 8M later). Having a 386 but not enough memory
seems a bit silly, I think. It seems most other software developers
have the same idea: OS/2, windows NT etc all want a /lot/ more that just
1M to run nicely.
But yes: if you are cramped for memory, coherent is certainly the way to
go - and likewise if you don't enjoy setting up and grappling with all
the problems of a new operating system (and believe me: people do have
problems with linux - it doesn't work on all machines, and setup can be
a pain if you aren't used to things like that). But I'd also like to
point out that for the price of coherent ($99?) you should have no
problems finding someone willing to sell you 2M of more memory, and use
linux happily (?) ever after.
>Please, if you have a priority list, put these three things on the top of the
>list:
>
> (1) rewrite the kernel so that RAM usage is cut under 500kB.
Well, we have different priorities: I have a nice enough machine, and I
want to get everything out of it. If that means I'll use bigger
programs, so be it: I don't consider the extra megabytes needed that
relevant. (and the source /is/ free, so somebody could rewrite it to
just use the minimum memory available, but I'm not interested to do it
myself).
> (2) contact someone who's known for developing high quality
> software tools (e.g. the producer of the Power Mix C compiler)
> and see if they won't participate in the major contribution to your
> public domain, with free software.
Somehow I don't see commercial software houses rushing in to develop
software for linux - but more importantly, I consider GNU software to be
among the best around. Yes, it's big, but it certainly is nice to have
around. Gcc-2.0 (already running under linux: in alpha-testing) has a
cc1-binary of about 800kB (so with make, gcc and cc1 all in memory at
once, you easily fill up >>1MB), but the code produced is well worth it.
I guess this comes as a shock to people using a compiler that fits in
64kB, but it's nice to have a development system that /really/ does a
good job.
> (3) write some good documentation in the implementation of Linux, as
> the written word of a designer's intent has value to a software
> developer far in excess to any value the source itself may have.
Yes: a valid point. I hate writing docs, and as this has been purely a
hacker project for my own enjoyment (started last april, and it's
finally getting useable indeed), I haven't written more than the
/absolute/ minimum required (and some people seem to think I didn't
write even that much :). Even the FAQ and the info-sheet were written
by others, I'm happy to say.
Linus