Xref: gmd.de comp.sys.next.advocacy:10238 comp.os.ms-windows.advocacy:10816
Path: gmd.de!xlink.net!howland.reston.ans.net!spool.mu.edu!
bloom-beacon.mit.edu!senator-bedfellow.mit.edu!athena.mit.edu!zmonster
From: zmon...@athena.mit.edu (Eric M Hermanson)
Newsgroups: comp.sys.next.advocacy,comp.os.ms-windows.advocacy
Subject: The Evil Empire has Ejaculated
Date: 27 Jul 1993 07:09:39 GMT
Organization: Massachusetts Institute of Technology
Lines: 44
Distribution: world
Message-ID: <232kbj$g5d@senator-bedfellow.MIT.EDU>
NNTP-Posting-Host: al-burro.mit.edu

The Evil Empire has ejaculated - NT has been "released".

(I eat my words from a posting I made earlier claiming NT is not a shipping
product, although I am halfway right - it is supposed to be on the shelves
at retail stores 3 weeks from now).

An internal Email from David Cutler (head of NT project for Microsoft).
This email can be found in comp.os.ms-windows.announce:
---------------------------------------------------------------------------
Subject: NT Release
Date: Monday, July 26, 1993 2:50PM

PSS has signed off and NT is offically released to manufacturing!

Let me state that again - NT is RELEASED!!!!!

This has been a long hard effort. You all have done a great job. NT
is really something to be proud of.

Thank you all for your contribution, especially in the final three
months where we managed to fix upwards of 200 bugs per day without
serious regessions. It took an unparalled effort on everyone's part
to accomplish this.


We met or exceeded all ship criteria!

d
-------------------------------------------------------------------

Upwards of 200 bug fixes a day for 3 MONTHS!?  That's 12 THOUSAND bugs,
apparantly fixed.  Well the March beta sucked, but after 12 THOUSAND bug fixes,
if NT isn't in good shape now, it'll never be!  8-)

I, for one, am glad they finally got the thing out the door.  Now NEXTSTEP
and NT can be compared, on a feature by feature basis.  No more excuses-
for EITHER side.  We can finally see how much better NEXTSTEP is - and with
that, I challenge my NT foes.

-Eric




Xref: gmd.de comp.sys.next.advocacy:10460 comp.unix.misc:6488
 comp.unix.sys5.r4:4382
Newsgroups: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!sol.ctr.columbia.edu!math.ohio-state.edu!uwm.edu!
cs.utexas.edu!uunet!psinntp!laidbak!binford!mcdchg!chinet!les
From: l...@chinet.chinet.com (Leslie Mikesell)
Subject: Re: NT goes to manufacturing
Message-ID: <CB298C.2wq@chinet.chinet.com>
Organization: Chinet Public Access UNIX
References: <232kbj$g5d@senator-bedfellow.MIT.EDU> <CAwvBq.58x@molly.uucp> 
<RANDY.93Jul30093730@dsndata.dsndata.com>
Date: Sun, 1 Aug 1993 03:19:24 GMT
Lines: 24

In article <RANDY.93J...@dsndata.dsndata.com>,

>From a developers standpoint on Linux; why wouldn't you choose a POSIX
>compliant OS (which Linux seems to be doing), that is FREE and
>supplied with source code, runs Windows 3.1 apps (which Linux soon
>will), and is 1/4 the resource HOG that SVR4 has become?

How about not being able to run any commercial apps?  Or get drivers
for equipment commonly run under Sysvr4 like intelligent ports boards?
If Linux were binary compatible or even had a binary translator to
convert stock elf/coff binaries and sysvr4 drivers into something usable
it would be a lot more useful. 

>For 1 reason alone: Freeing yourself from your dependency on the OS
>vendor's ability to find and fix bugs is reason enough for me.

I'd read that as not being able to take advantage of a vendor's
ability to find and fix bugs...  But, I'd give it a try if there
were any chance it would be useful in my environment which won't
happen until there is a Novell or LanMan server that will run on
it.

Les Mikesell
  l...@chinet.com




Newsgroups: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!Germany.EU.net!incom!odb!eurom!misch
From: mi...@eurom.fsag.rhein-main.de (Michaela Merz)
Subject: Re: NT goes to manufacturing
Content-Type: text/plain; charset=iso-8859-1
Message-ID: <CB2zKp.HB3@eurom.fsag.rhein-main.de>
Followup-To: comp.unix.sys5.r4
Content-Transfer-Encoding: 8bit
Organization: Godcorp
X-Newsreader: MINEWS [FSAG] Version: 0.1
Date: Sun, 1 Aug 1993 12:48:24 GMT
Lines: 28


On 31 Jul 93 15:36:47 GMT, d...@mtgzfs3.att.com (David E. Wexelblat) wrote:


> Linux is a toy.  When it stop changing every single day, perhaps it will
> be considered a viable non-hacker platform.  But no one could seriously
> use Linux in a production environment of any size.  SVR4, OS/2, NT, and
> Solaris will be used in such environments.
> 
> Why is it that Linux users have an over-inflated sense of the importance
> of their OS?

> --
> David Wexelblat <d...@mtgzfs3.att.com>  (908) 957-5871  Fax: (908) 957-5305
> AT&T Bell Laboratories, 200 Laurel Ave - 3F-428, Middletown, NJ  07748

what else should you say? We're informed, that big software companies trying
to keep LINUX as down as possible. Linux is the first real choices to cut
the chains and become free. We will do everything possible to get Linux 
into business. Linux does have some bugs - what about at&t unix? If I
find something wrong, I'd rather want to fix it than to wait half a year
for somedbody who is allowed to take a look into the at&t source codes -
or to pay a lot of money for a so called 'up -date/grade'. What do you
think, how many licenses did at&t allready lost because of Linux?

Michaela

 *** eurom: Das Mailboxsystem der Free Software Association *** 




Newsgroups: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!sol.ctr.columbia.edu!spool.mu.edu!olivea!uunet!
sparky!dsndata!backbone!backbone!wayne
From: wa...@backbone.uucp (Wayne Schlitt)
Subject: Re: NT goes to manufacturing
In-Reply-To: dwex@mtgzfs3.att.com's message of 31 Jul 93 15: 36:47 GMT
Message-ID: <WAYNE.93Aug1101741@backbone.uucp>
Sender: wayne@backbone (Wayne Schlitt)
Organization: The Backbone Cabal
References: <232kbj$g5d@senator-bedfellow.MIT.EDU> <CAwvBq.58x@molly.uucp>
	<RANDY.93Jul30093730@dsndata.dsndata.com>
	<CB1CpC.GDJ@cbnewsj.cb.att.com>
Date: Sun, 1 Aug 1993 16:17:41 GMT
Lines: 126

In article <CB1CpC....@cbnewsj.cb.att.com> d...@mtgzfs3.att.com 
(David E. Wexelblat) writes:
> In article <RANDY.93Jul30093...@dsndata.dsndata.com> ran...@cse.unl.edu 
writes:
> > >>>>> "Jim" == Jim Vlcek <uunet!molly!vlcek> writes:
> >   Jim> Suggestions:
> > 
> >   Jim> - UNIX weenies, rally yourselves around SVR4.  For God's sake,
> >   Jim> forget about BSD and Linux - even for your hobby projects.
> > 
> > I would agree with most of your post with the exception of the above.
> > 
> > My experience with SVR4 has shown that it has more bugs and
> > shortcomings than many other UNIX variants I have had experience with
> > over the years.
> > 
> 
> Please.  I've been running SVR4 for two years.   I find it to be compatible
> with just about everything - I almost never have to change code to make it
> work on SVR4, whether the code was written for BSD or for SVR3 (the exception
> is often conversion to POSIX signals or ttys).  It run SCO ODT binaries, it
> runs DOS (and SVR4.2 runs Windows).

I have been running SVR4 for two years also.  I find that just about
everything off the net needs at least a little tweeking, and sometimes
a _lot_ of tweeking to get it to run.  

> 
> > From a developers standpoint on Linux; why wouldn't you choose a POSIX
> > compliant OS (which Linux seems to be doing), that is FREE and
> > supplied with source code, runs Windows 3.1 apps (which Linux soon
> > will), and is 1/4 the resource HOG that SVR4 has become?
> > 
> 
> SVR4.2 will run in just as small a system as any reasonably-usable Linux
> system.  It runs Windows.  It also support NetWare (if you get UnixWare).
> Linux will not run Windows any time soon (the WABI-type stuff).  That's
> not something that gets thrown together overnight.  SVR4 does networking
> better than Linux, and has commercial application support.

I can not verify the first statement, and my gut feel says that it is
wrong.  I regularly see people posting in the linux group that they
are running X in 4MB of memory.  I get the feeling that they don't run
many programs, and it's not blazingly fast, but I doubt that SVR4
would do better...

It is also true that a WABI-type system doesn't get thrown together
over night, but I am willing to be that there will be a usable system
ready for Linux within the next year.  Things like this often take
a _very_ long time from comercial software providers too.  

> 
> Linux is a toy.  When it stop changing every single day, perhaps it will
> be considered a viable non-hacker platform.  But no one could seriously
> use Linux in a production environment of any size.  SVR4, OS/2, NT, and
> Solaris will be used in such environments.

hmmm... 6-8 years ago, I remember hearing very similar things about
GNU-emacs vs UNIPRESS-emacs.  4-5 years ago, people were saying that
GCC would never be able to keep up or compete with commercially produce
C compilers. 

The thing you got to remember is that the ability to _try_ the product
with very little cost (it only cost time, and that doesn't go through
the purchasing department) is a big plus.  If the product is
reasonably high quality, then you tend to not go looking any
further... 

The other _really_ big plus is that you can get bug fixes to _your
particular problem_ off the net quickly, and if no such bug fix
exists, you can make your own fix.  These bug fixes care often
incorporated into the next release, so you don't have to fix them
every time.  Even if you don't create a bug fix, being able to look at
the source code can help you find and fix things _much_ quicker.

To the commercial software developer, or large system administrator,
having source and being able to fix things your self is a _very_ major
item.  Nothing is more frustrating that finding a critical problem
(like SVR4's nfs locking) and being told that there is no work around,
and no fix planned.  Worse, it is close to impossible for you to fix
it your self.  Even if the bug _is_ fixed, you rarely get it before
the next major OS release, or often the release after that because you
have missed the dead line for QA and manufacturing...


> > For 1 reason alone: Freeing yourself from your dependency on the OS
> > vendor's ability to find and fix bugs is reason enough for me.
> > 
> 
> For you.  Irrelevant in the commercial arena.  

bunk.  complete bunk.


> 
> Why is it that Linux users have an over-inflated sense of the importance
> of their OS?
> 

Well, probably because the user base is growing so quickly.




1.5 years ago, I bought Dell Unix SVR4 (issue 2.1) for my home
machine.  Today, I would go with Linux.  The work that I have done to
write device drivers and fix problems with SVR4 will never be seen by
most other people because Dell and USL don't care about my code/fixes
(NIH).  Linux people seem to be glad to get any help.  When you add up
all the free programming that people will do for public available
systems, the commercial people will have to have a huge staff just to
keep up.

Free software doesn't work for all areas, but for operating system
tools, and now for operating systems them selves, it seems that it can
work very well.  

I have noticed a distinct drop in volume/interest in the these groups
in the last year or so.  People who used to be rallying around Dell,
or Univel are now over in the Linux of 386bsd groups.

No, neither Linux nor 386bsd is ready for prime time yet, but don't
count them out too quickly.  In the long run, they are going to be
_very_ hard to stop.


-wayne

Path: gmd.de!xlink.net!sol.ctr.columbia.edu!math.ohio-state.edu!
darwin.sura.net!gatech!rutgers!att-out!cbnewsj!dwex
From: d...@mtgzfs3.att.com (David E. Wexelblat)
Newsgroups: comp.unix.sys5.r4
Subject: Re: NT goes to manufacturing
Message-ID: <CB3LFK.CpD@cbnewsj.cb.att.com>
Date: 1 Aug 93 20:40:30 GMT
References: <CB2zKp.HB3@eurom.fsag.rhein-main.de>
Sender: n...@cbnewsj.cb.att.com (NetNews Administrator)
Organization: AT&T Bell Laboratories
Lines: 52
Nntp-Posting-Host: mtgzfs5.gaz.att.com

In article <CB2zKp....@eurom.fsag.rhein-main.de> 
mi...@eurom.fsag.rhein-main.de (Michaela Merz) writes:
> 
> On 31 Jul 93 15:36:47 GMT, d...@mtgzfs3.att.com (David E. Wexelblat) wrote:
> 
> 
> > Linux is a toy.  When it stop changing every single day, perhaps it will
> > be considered a viable non-hacker platform.  But no one could seriously
> > use Linux in a production environment of any size.  SVR4, OS/2, NT, and
> > Solaris will be used in such environments.
> > 
> > Why is it that Linux users have an over-inflated sense of the importance
> > of their OS?
> 
> > --
> > David Wexelblat <d...@mtgzfs3.att.com>  (908) 957-5871  Fax: (908) 957-5305
> > AT&T Bell Laboratories, 200 Laurel Ave - 3F-428, Middletown, NJ  07748
> 
> what else should you say? We're informed, that big software companies trying
> to keep LINUX as down as possible. Linux is the first real choices to cut
> the chains and become free. We will do everything possible to get Linux 
> into business. Linux does have some bugs - what about at&t unix? If I
> find something wrong, I'd rather want to fix it than to wait half a year
> for somedbody who is allowed to take a look into the at&t source codes -
> or to pay a lot of money for a so called 'up -date/grade'. What do you
> think, how many licenses did at&t allready lost because of Linux?
> 

1) AT&T doesn't own Unix - Novell does.
2) Even when AT&T did own Unix, it was the USL subsidiary.
3) I have no and never had any connection to USL.  AT&T is a company of
   300,000 employees.
4) AT&T hasn't lost anything to Linux, because AT&T doesn't own Unix.

I coordinate XFree86 - you know, that little FreeWare project that you
all depend on?  I run SVR4, by choice.  I wouldn't even consider using
Linux at home, let alone somewhere that really mattered.

How many of your company's dollars are you going to invest in HackerWare?

> Michaela
> 
>  *** eurom: Das Mailboxsystem der Free Software Association *** 


--
David Wexelblat <d...@mtgzfs3.att.com>  (908) 957-5871  Fax: (908) 957-5305
AT&T Bell Laboratories, 200 Laurel Ave - 3F-428, Middletown, NJ  07748

XFree86 requests should be addressed to <xfre...@physics.su.oz.au>

"Shining, flying, purple wolfhound, show me where you are."
	Yes, "Yours Is No Disgrace"

Xref: gmd.de comp.sys.next.advocacy:10463 comp.unix.misc:6490 
comp.unix.sys5.r4:4389
Newsgroups: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!sol.ctr.columbia.edu!spool.mu.edu!olivea!uunet!
sparky!dsndata!backbone!backbone!wayne
From: wayne@backbone.uucp (Wayne Schlitt)
Subject: Re: NT goes to manufacturing
In-Reply-To: les@chinet.chinet.com's message of Sun, 1 Aug 1993 03: 19:24 GMT
Message-ID: <WAYNE.93Aug1202432@backbone.uucp>
Sender: wayne@backbone (Wayne Schlitt)
Organization: The Backbone Cabal
References: <232kbj$g5d@senator-bedfellow.MIT.EDU> <CAwvBq.58x@molly.uucp>
	<RANDY.93Jul30093730@dsndata.dsndata.com>
	<CB298C.2wq@chinet.chinet.com>
Date: Mon, 2 Aug 1993 02:24:32 GMT
Lines: 69

In article <CB298...@chinet.chinet.com> 
l...@chinet.chinet.com (Leslie Mikesell) writes:
> In article <RANDY.93J...@dsndata.dsndata.com>,
> >From a developers standpoint on Linux; why wouldn't you choose a POSIX
> >compliant OS (which Linux seems to be doing), that is FREE and
> >supplied with source code, runs Windows 3.1 apps (which Linux soon
> >will), and is 1/4 the resource HOG that SVR4 has become?
> 
> How about not being able to run any commercial apps?  Or get drivers
> for equipment commonly run under Sysvr4 like intelligent ports boards?

Availability of commercial apps is more of a concern for users than
developers, and there is _at least_ one commercial app available for
Linux, but your point is well taken.  It is also a point that is well
understood by most of the Linux folks and they are working on it.


Availability of device drivers on the other hand is a much more of a
gray area.  It is true that Linux lacks device drivers for a lot of
the "intelligent" multi-port I/O boards, but then, it supports a lot of
devices that most SVR4's don't support, such as the sound blasters and
various CD-ROMS.  I saw drivers for the adaptec 1542C cards for 386bsd
and Linux before I saw them for any SVR4's.


> If Linux were binary compatible or even had a binary translator to
> convert stock elf/coff binaries and sysvr4 drivers into something usable
> it would be a lot more useful. 

Yes, and there are people working on this area.  Don't expect miracles
soon, but this is one angle that the Linux folks are hoping to get
commercial apps running under Linux.  (The WABI project, writing their
own apps, and convincing developers to port to Linux are other areas
that they are working on...)

> 
> >For 1 reason alone: Freeing yourself from your dependency on the OS
> >vendor's ability to find and fix bugs is reason enough for me.
> 
> I'd read that as not being able to take advantage of a vendor's
> ability to find and fix bugs...

If by this, you mean that you can't get other people to find/fix bugs
for you, or help you with your problems, then you are wrong.  There is
a list of companies/people who will support Linux, for a fee.  (No, I
don't expect the support to be great, but then, this is kind of early
in the game for Linux...)

>                                  But, I'd give it a try if there
> were any chance it would be useful in my environment which won't
> happen until there is a Novell or LanMan server that will run on
> it.


No one said that Linux (especially in it's current form) is useful to
all people in all situations.  I hope that no one tries to force you
to use any system when it is not right for the job.  But, when someone
says that "UNIX weenies, [should] rally yourselves around SVR4.  For
God's sake, forget about BSD and Linux - even for your hobby
projects.", then you should not let such comments go unanswered.

Linux has some very real strengths that SVR4 does not have, and it
appears that Linux has a better chance of fixing it's weaknesses than
SVR4 has of matching Linux's strengths.  (I.e., if you could get a
full SVR4 for $60, with full source, then Linux would be in serious
trouble, but I don't see that happening any time soon....)



-wayne

Newsgroups: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!Germany.EU.net!news.dfn.de!xlink.net!sol.ctr.columbia.edu!
math.ohio-state.edu!wupost!uunet!tarpit!gator!not-for-mail
From: la...@gator.oau.org (Larry D Snyder)
Subject: Re: NT goes to manufacturing
Message-ID: <CB4or4.3Bt@gator.oau.org>
Organization: Gator Communications, Orlando, Florida
References: <232kbj$g5d@senator-bedfellow.MIT.EDU> <CAwvBq.58x@molly.uucp> 
<RANDY.93Jul30093730@dsndata.dsndata.com> <CB1CpC.GDJ@cbnewsj.cb.att.com> 
<WAYNE.93Aug1101741@backbone.uucp>
Date: Mon, 2 Aug 1993 10:49:52 GMT
Lines: 64

wa...@backbone.uucp (Wayne Schlitt) writes:

>> SVR4.2 will run in just as small a system as any reasonably-usable Linux
>> system.  It runs Windows.  It also support NetWare (if you get UnixWare).
>> Linux will not run Windows any time soon (the WABI-type stuff).  That's
>> not something that gets thrown together overnight.  SVR4 does networking
>> better than Linux, and has commercial application support.

>I can not verify the first statement, and my gut feel says that it is
>wrong.  I regularly see people posting in the linux group that they
>are running X in 4MB of memory.  I get the feeling that they don't run
>many programs, and it's not blazingly fast, but I doubt that SVR4
>would do better...

You might be impressed with current SVR4 performance

>> Linux is a toy.  When it stop changing every single day, perhaps it will
>> be considered a viable non-hacker platform.  But no one could seriously
>> use Linux in a production environment of any size.  SVR4, OS/2, NT, and
>> Solaris will be used in such environments.

>hmmm... 6-8 years ago, I remember hearing very similar things about
>GNU-emacs vs UNIPRESS-emacs.  4-5 years ago, people were saying that
>GCC would never be able to keep up or compete with commercially produce
>C compilers. 

>The thing you got to remember is that the ability to _try_ the product
>with very little cost (it only cost time, and that doesn't go through
>the purchasing department) is a big plus.  If the product is
>reasonably high quality, then you tend to not go looking any
>further... 

How can one do develpment for products on an operating such as Linux
for real world applications.  Linux is like Minix -- designed for hobbiest
and educational markets..

>The other _really_ big plus is that you can get bug fixes to _your
>particular problem_ off the net quickly, and if no such bug fix
>exists, you can make your own fix.  These bug fixes care often
>incorporated into the next release, so you don't have to fix them
>every time.  Even if you don't create a bug fix, being able to look at
>the source code can help you find and fix things _much_ quicker.

Most of us are busy running our systems to support real world applications
and have no desire to go tracing kernel bugs.


>> Why is it that Linux users have an over-inflated sense of the importance
>> of their OS?

>Well, probably because the user base is growing so quickly.

It's like Amiga users verses PC-DOS users, has anyone else noticed?

>I have noticed a distinct drop in volume/interest in the these groups
>in the last year or so.  People who used to be rallying around Dell,
>or Univel are now over in the Linux of 386bsd groups.

I seriously doubt if commercial ventures have switched from Dell or
Univel to Linux..

-- 
Larry Snyder                                    Internet: la...@gator.oau.org
Orlando, Florida                            UUCP: ..!uunet!tarpit!gator!larry

Xref: gmd.de comp.sys.next.advocacy:10467 comp.unix.misc:6496 
comp.unix.sys5.r4:4393
Newsgroups: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!sol.ctr.columbia.edu!math.ohio-state.edu!
cs.utexas.edu!uunet!tarpit!gator!not-for-mail
From: la...@gator.oau.org (Larry D Snyder)
Subject: Re: NT goes to manufacturing
Message-ID: <CB4oto.3F9@gator.oau.org>
Organization: Gator Communications, Orlando, Florida
References: <232kbj$g5d@senator-bedfellow.MIT.EDU> <CAwvBq.58x@molly.uucp> 
<RANDY.93Jul30093730@dsndata.dsndata.com> 
<CB298C.2wq@chinet.chinet.com> <WAYNE.93Aug1202432@backbone.uucp>
Date: Mon, 2 Aug 1993 10:51:23 GMT
Lines: 16

wayne@backbone.uucp (Wayne Schlitt) writes:

>Availability of commercial apps is more of a concern for users than
>developers, and there is _at least_ one commercial app available for
>Linux, but your point is well taken.  It is also a point that is well
>understood by most of the Linux folks and they are working on it.

how about drivers?  Are digiboard drivers available for linux?
How about the selection of hardware add-on products which exist
for other variations of Unix?  I like Word Perfect for writing,
how about WP for Linux -- fat chance.


-- 
Larry Snyder                                    Internet: la...@gator.oau.org
Orlando, Florida                            UUCP: ..!uunet!tarpit!gator!larry

Newsgroups: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!math.fu-berlin.de!easix!flyer!flatlin!bad
From: b...@flatlin.ka.sub.org (Christoph Badura)
Subject: Re: NT goes to manufacturing
Organization: Guru Systems/Funware Department
Date: Mon, 2 Aug 1993 15:31:33 GMT
Message-ID: <CB51sL.Bou@flatlin.ka.sub.org>
References: <232kbj$g5d@senator-bedfellow.MIT.EDU> <CAwvBq.58x@molly.uucp> 
<RANDY.93Jul30093730@dsndata.dsndata.com> <CB1CpC.GDJ@cbnewsj.cb.att.com> 
<WAYNE.93Aug1101741@backbone.uucp> <CB4or4.3Bt@gator.oau.org>
Lines: 40

In <CB4or4....@gator.oau.org> la...@gator.oau.org (Larry D Snyder) writes:

>wa...@backbone.uucp (Wayne Schlitt) writes:
>>I regularly see people posting in the linux group that they
>>are running X in 4MB of memory.  I get the feeling that they don't run
>>many programs, and it's not blazingly fast, but I doubt that SVR4
>>would do better...
>You might be impressed with current SVR4 performance

Unlikely.  Under no circumstance is SVR4 going to perform better than
Linux with 4MB RAM.  Not to mention the abysmal network performance.

>How can one do develpment for products on an operating such as Linux
>for real world applications.

And why not?  You can use a reasonably standard curses library under
Linux if you want to develop character based applications.  You can
use standard X windows and Motif libraries under Linux.  You have
POSIX conforming C library.

Tell me, what else do you need to develop applications under UNIX.  Or
was that the usual FUD one has come to expect from you?

>Most of us are busy running our systems to support real world applications
>and have no desire to go tracing kernel bugs.

Well, if you're busy keeping you system running, I suggest you take a
look into systems that don't require that amount of handholding.  Not
that I think that Linux is a candidate for such a system.

>I seriously doubt if commercial ventures have switched from Dell or
>Univel to Linux..

Well, you would be amazed to hear how many are evaluating Linux as an
alternative for certain applications.
-- 
    Christoph Badura  ---  b...@flatlin.ka.sub.org  ---  +49 721 606137

Personally, I don't care whether someone is cool enough to quote Doug
Gwyn--I only care whether Doug Gwyn is cool enough to quote. -- Larry Wall

Newsgroups: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!news.dfn.de!darwin.sura.net!europa.eng.gtefsd.com!
uunet!wyvern!taylor.wyvern.com!mark
From: m...@taylor.uucp (Mark A. Davis)
Subject: Re: NT goes to manufacturing
Organization: Lake Taylor Hospital Computer Services
Date: Mon, 02 Aug 1993 18:04:39 GMT
Message-ID: <1993Aug02.180439.11053@taylor.uucp>
References: <232kbj$g5d@senator-bedfellow.MIT.EDU> <CAwvBq.58x@molly.uucp> 
<RANDY.93Jul30093730@dsndata.dsndata.com> <CB1CpC.GDJ@cbnewsj.cb.att.com> 
<WAYNE.93Aug1101741@backbone.uucp> <CB4or4.3Bt@gator.oau.org> 
<CB51sL.Bou@flatlin.ka.sub.org>
Lines: 49

b...@flatlin.ka.sub.org (Christoph Badura) writes:

>In <CB4or4....@gator.oau.org> la...@gator.oau.org (Larry D Snyder) writes:

>>wa...@backbone.uucp (Wayne Schlitt) writes:
>>>I regularly see people posting in the linux group that they
>>>are running X in 4MB of memory.  I get the feeling that they don't run
>>>many programs, and it's not blazingly fast, but I doubt that SVR4
>>>would do better...
>>You might be impressed with current SVR4 performance

>Unlikely.  Under no circumstance is SVR4 going to perform better than
>Linux with 4MB RAM.  Not to mention the abysmal network performance.

>>How can one do develpment for products on an operating such as Linux
>>for real world applications.

>And why not?  You can use a reasonably standard curses library under
>Linux if you want to develop character based applications.  You can
>use standard X windows and Motif libraries under Linux.  You have
>POSIX conforming C library.

>Tell me, what else do you need to develop applications under UNIX.  Or
>was that the usual FUD one has come to expect from you?

You need a STABLE version to come out.... a software company cannot afford
the R&D&porting time to change their application every few months to keep
up with the OS changing all the time.  This is the main problem.  Imagine
the distribution problems, countless upgrades, etc.  Linux will someday
reach that point, but it simply is not there yet,

IF Linux had COFF (or complete ELF) capabilities, then it would be able run
existing commercial applications.  

>>I seriously doubt if commercial ventures have switched from Dell or
>>Univel to Linux..

>Well, you would be amazed to hear how many are evaluating Linux as an
>alternative for certain applications.

I think it would be a good idea, as long as porting is easy and the constant
changes don't continuosly outdate their porting efforts.  Still, COFF
seems like a much easier alternative (it's something the Linux world CAN
control)
-- 
  /--------------------------------------------------------------------------\
  | Mark A. Davis    | Lake Taylor Hospital | Norfolk, VA (804)-461-5001x431 |
  | Sys.Administrator|  Computer Services   | m...@taylor.wyvern.com   .uucp |
  \--------------------------------------------------------------------------/

Xref: gmd.de comp.sys.next.advocacy:10486 comp.unix.misc:6510 
comp.unix.sys5.r4:4415
Newsgroups: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!math.fu-berlin.de!news.netmbx.de!Germany.EU.net!
mcsun!uunet!spool.mu.edu!uwm.edu!wupost!waikato!comp.vuw.ac.nz!
ferrari.datamark.co.nz!thomas
From: tho...@datamark.co.nz (Thomas Beagle)
Subject: Re: NT goes to manufacturing
Message-ID: <1993Aug2.204307.26565@datamark.co.nz>
Date: Mon, 2 Aug 1993 20:43:07 GMT
References: <CB298C.2wq@chinet.chinet.com>> <WAYNE.93Aug1202432@backbone.uucp> 
<CB4oto.3F9@gator.oau.org>
Organization: Datamark International Ltd.
Lines: 19

In article <CB4ot...@gator.oau.org> la...@gator.oau.org (Larry D Snyder) writes:
>how about drivers?  Are digiboard drivers available for linux?
>How about the selection of hardware add-on products which exist
>for other variations of Unix?  I like Word Perfect for writing,
>how about WP for Linux -- fat chance.

Well, consider the crappy device support that Unixware has. This is a
commercial release.

I've got the application server install pack sitting here but I can't
install because it only supports Adaptec and DPT SCSI controllers.

And their video card support is a lot worse than that of Linux. (Which
piggybacks on XFree86 (or whatever it's called)).

-- 
   Thomas Beagle | tho...@datamark.co.nz Work:64 4 2338186 Home:64 4 4993832
Technical Writer |        Why is it that I seemed to be fated
  Wellington, NZ |            to live my life vicariously?

Newsgroups: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!Germany.EU.net!mcsun!uunet!olivea!pagesat!decwrl!csus.edu!
netcom.com!meb
From: m...@netcom.com (Michael Burg)
Subject: Re: NT goes to manufacturing
Message-ID: <mebCB6E6u.JB2@netcom.com>
Followup-To: comp.sys.next.advocacy,comp.unix.misc,comp.unix.sys5.r4
Organization: NETCOM On-line Communication Services (408 241-9760 guest)
X-Newsreader: TIN [version 1.1 PL8]
References: <1993Aug02.180439.11053@taylor.uucp>
Date: Tue, 3 Aug 1993 08:56:53 GMT
Lines: 65

Mark A. Davis (m...@taylor.uucp) wrote:
: b...@flatlin.ka.sub.org (Christoph Badura) writes:

: >In <CB4or4....@gator.oau.org> la...@gator.oau.org (Larry D Snyder) writes:

: >Unlikely.  Under no circumstance is SVR4 going to perform better than
: >Linux with 4MB RAM.  Not to mention the abysmal network performance.

There maybe a exception in the filesystem areas. SVR4.2 is now
using the Veritas Filesystem (or at least the beta I was running). This
is a logging filesystem with very agresive read-ahead. The filesystem
will increase the size of the read-ahead for each sequential read each time up
until a 64K read-ahead block size is reached. This turns out some impressive
numbers. A Maxtor LXT-540S with a UFS filesystem was ~ 800-900K on a 80mb
file. Under Vertias FS, this was about 1100K to 1300K. (This was on a 486DX/50
with 16MB and a Adaptec 1542B).

The time to perform a fsck is on the order of 1 second per gigabyte
(not megabyte) assuming the filesystem log is 'somewhat' intact.

Also, you can vary the levels of filesystem 'sync'ing to boost
performance at the expense of a more corrupt filesystem if a crash
occurred. (You could turn syncing off completely to create fast /tmp
filesystems, the buffers would only be flushed when fsflush woke up - this
included inode updates)

This functionality could be added to Linux.. *hint* *hint* *hint*

: You need a STABLE version to come out.... a software company cannot afford
: the R&D&porting time to change their application every few months to keep
: up with the OS changing all the time.  This is the main problem.  Imagine
: the distribution problems, countless upgrades, etc.  Linux will someday
: reach that point, but it simply is not there yet,

Also keep in mind, a lot of application vendors would be unwilling to port
to 'new' O/S that would have a small offerring (at least initially). In fact,
some might want ca$h to do it!

: IF Linux had COFF (or complete ELF) capabilities, then it would be able run
: existing commercial applications.  

Except for SCO Unix, COFF is basicly dead in the SVR4 world - remember that
all binaries compiled under SVR4 are ELF.

This could be a VERY big undertaking. Think of all thoses things the Linux O/S
would have to support in order to run a SVR4.0 ABI program. Here's a
few I can think of.

- Most of the SVR4 system calls. (SVR4.2 is worse, you have the Enhanced
Security syscalls to worry about. And in the future there will be the
Light Weight Process, i.e. threads, system calls to worry about..
SVR4/ESMP, the multiprocessor version of SVR4, has this now)
- A emulation (not necessarily a recreation) of how SVR4 does networking
  with TLI and STREAMS modules.
- ELF support: Libraries (libc.so - a large undertaking!), Run-Time linker
  support - ELF style, and the system calls to support it.

It doesn't mean it can't be done.. Just a _lot_ of work to do there.
(And I fear, Linux would be just as BIG as SVR4 is now)

-- Michael
m...@netcom.com
(used to be m...@smsc.sony.com)

Newsgroups: comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!math.fu-berlin.de!news.netmbx.de!Germany.EU.net!
mcsun!uunet!wyvern!taylor.wyvern.com!mark
From: m...@taylor.uucp (Mark A. Davis)
Subject: RE: Linux COFF (WAS: NT goes)
Organization: Lake Taylor Hospital Computer Services
Date: Tue, 03 Aug 1993 14:38:25 GMT
Message-ID: <1993Aug03.143825.113@taylor.uucp>
References: <1993Aug02.180439.11053@taylor.uucp> <mebCB6E6u.JB2@netcom.com>
Lines: 48

m...@netcom.com (Michael Burg) writes:

>Mark A. Davis (m...@taylor.uucp) wrote:
>: You need a STABLE version to come out.... a software company cannot afford
>: the R&D&porting time to change their application every few months to keep
>: up with the OS changing all the time.  This is the main problem.  Imagine
>: the distribution problems, countless upgrades, etc.  Linux will someday
>: reach that point, but it simply is not there yet,

>Also keep in mind, a lot of application vendors would be unwilling to port
>to 'new' O/S that would have a small offerring (at least initially). In fact,
>some might want ca$h to do it!

>: IF Linux had COFF (or complete ELF) capabilities, then it would be able run
>: existing commercial applications.  

>Except for SCO Unix, COFF is basicly dead in the SVR4 world - remember that
>all binaries compiled under SVR4 are ELF.

Not true.  TONS and TONS of business COFF stuff is run under SVR4 which has
not been or will not be ported any time soon.  Lots of businesses are quite
content with SCO, and those companies writing software for them will end
up with COFF.  I doubt sales of SVR4 would have amounted to much if they HADN'T
supported COFF binaries; because those seeking to upgrade may not have been
able to run the software which they depend on.

>This could be a VERY big undertaking. Think of all thoses things the Linux O/S
>would have to support in order to run a SVR4.0 ABI program. Here's a
>few I can think of.

Not defending the COFF or ELF standards, just trying to put a real-world
twist on what they mean to those who buy computers.  We need applications!
EVERY UNIX needs applications!!!  Technology is great, Unix is great, but
application software gets the work done.

[...]

>It doesn't mean it can't be done.. Just a _lot_ of work to do there.
>(And I fear, Linux would be just as BIG as SVR4 is now)

It might be.  But if that is what it takes to bring real applications to
it, and it is still free and flexible, more power to it!

-- 
  /--------------------------------------------------------------------------\
  | Mark A. Davis    | Lake Taylor Hospital | Norfolk, VA (804)-461-5001x431 |
  | Sys.Administrator|  Computer Services   | m...@taylor.wyvern.com   .uucp |
  \--------------------------------------------------------------------------/

Newsgroups: comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!math.fu-berlin.de!easix!flyer!flatlin!bad
From: b...@flatlin.ka.sub.org (Christoph Badura)
Subject: Re: Linux COFF (WAS: NT goes)
Organization: Guru Systems/Funware Department
Date: Wed, 4 Aug 1993 14:23:31 GMT
Message-ID: <CB8nz8.1D0@flatlin.ka.sub.org>
References: <1993Aug02.180439.11053@taylor.uucp> <mebCB6E6u.JB2@netcom.com> 
<1993Aug03.143825.113@taylor.uucp>
Lines: 42

In <1993Aug03.143825....@taylor.uucp> m...@taylor.uucp (Mark A. Davis) writes:

>m...@netcom.com (Michael Burg) writes:
>>Except for SCO Unix, COFF is basicly dead in the SVR4 world - remember that
>>all binaries compiled under SVR4 are ELF.

>Not true.

I suppose you're questioning the first half of Michael's statement and
not the latter.

Anyway I must say that Michael has a point.  No SVR4 vendor who
doesn't also own a SVR3 source licence is not in the position to fix
any bugs in the R3 shared library binaries that are delivered with the
SVR4 source.  The SVR4 source comes with neither the source for the R3
shared libraries nor the tools to create a R3 library to boot.

Considering this I think it's safe to say that COFF is deprecated
under SVR4 and that it is strictly a vehicle to get applications
running on it until they have been ported to SVR4.

>TONS and TONS of business COFF stuff is run under SVR4 which has
>not been or will not be ported any time soon.

This, however, is true too.

>>[supporting ELF or COFF under Linux.]
>>It doesn't mean it can't be done.. Just a _lot_ of work to do there.
>>(And I fear, Linux would be just as BIG as SVR4 is now)
>It might be.  But if that is what it takes to bring real applications to
>it, and it is still free and flexible, more power to it!

I doubt that.  The kernel support required for ELF and COFF isn't that
large.  And believe it is possible, if painful, to change the Linux
source in a way that the ioctl codes and system call parameters are
binary compatible with SysV/386.

-- 
    Christoph Badura  ---  b...@flatlin.ka.sub.org  ---  +49 721 606137

Personally, I don't care whether someone is cool enough to quote Doug
Gwyn--I only care whether Doug Gwyn is cool enough to quote. -- Larry Wall

Newsgroups: comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!Germany.EU.net!mcsun!uunet!europa.eng.gtefsd.com!darwin.sura.net!
news-feed-2.peachnet.edu!umn.edu!uum1!gw.digibd.com!digibd!rick
From: r...@digibd.digibd.com (Rick Richardson)
Subject: Re: Linux COFF (WAS: NT goes)
Message-ID: <rick.744526681@digibd>
Sender: n...@gw.digibd.com (USENET News)
Nntp-Posting-Host: digibd.digibd.com
Organization: DigiBoard, Incorporated, Eden Prairie,MN
References: <1993Aug02.180439.11053@taylor.uucp> <mebCB6E6u.JB2@netcom.com> 
<1993Aug03.143825.113@taylor.uucp> <CB8nz8.1D0@flatlin.ka.sub.org>
Date: Thu, 5 Aug 1993 04:58:01 GMT
Lines: 35

b...@flatlin.ka.sub.org (Christoph Badura) writes:

>In <1993Aug03.143825....@taylor.uucp> m...@taylor.uucp (Mark A. Davis) writes:

>>m...@netcom.com (Michael Burg) writes:
>>>Except for SCO Unix, COFF is basicly dead in the SVR4 world - remember that
>>>all binaries compiled under SVR4 are ELF.

>>Not true.

>I suppose you're questioning the first half of Michael's statement and
>not the latter.

I'd even question the latter.  I happen to use the ISC 3.0 compiler
*every* day, hosted on a UnixWare box, to generate maximally portable
COFF binaries.  No, I've never compiled them with shared libraries -
I've always felt that the static linking in SVR3.2 shared libraries
was too fragile to trust.

>>TONS and TONS of business COFF stuff is run under SVR4 which has
>>not been or will not be ported any time soon.

>This, however, is true too.

And from where I sit, every Xenix, SCO, and SVR3.2 system on the
planet will have to vaporize before I'll switch to ELF (except
for X binaries).  Its simply a matter of distributors not wanting
to stock 17 versions of the same program.

-Rick
--
Rick Richardson		Senior Staff Engineer	Writing X Widgets?
DigiBoard, Inc.		Email: r...@digibd.com	Get Bert Bos's 'wbuild 2.0'
6400 Flying Cloud Dr.	Fax:   (612) 943-0803	Any other way is primitive.
Eden Prarie, MN 55344	Tel:   (612) 943-5383	<standard disclaimer>

Newsgroups: comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!sol.ctr.columbia.edu!math.ohio-state.edu!sdd.hp.com!
swrinde!news.dell.com!tadpole.com!uunet!wyvern!taylor.wyvern.com!mark
From: m...@taylor.uucp (Mark A. Davis)
Subject: Re: Linux COFF (WAS: NT goes)
Organization: Lake Taylor Hospital Computer Services
Date: Thu, 05 Aug 1993 20:41:48 GMT
Message-ID: <1993Aug05.204148.7419@taylor.uucp>
References: <1993Aug02.180439.11053@taylor.uucp> <mebCB6E6u.JB2@netcom.com> 
<1993Aug03.143825.113@taylor.uucp> <CB8nz8.1D0@flatlin.ka.sub.org> 
<rick.744526681@digibd>
Lines: 32

r...@digibd.digibd.com (Rick Richardson) writes:

>b...@flatlin.ka.sub.org (Christoph Badura) writes:

>>m...@taylor.uucp (Mark A. Davis) writes:

>>>m...@netcom.com (Michael Burg) writes:

>>>TONS and TONS of business COFF stuff is run under SVR4 which has
>>>not been or will not be ported any time soon.

>>This, however, is true too.

>And from where I sit, every Xenix, SCO, and SVR3.2 system on the
>planet will have to vaporize before I'll switch to ELF (except
>for X binaries).  Its simply a matter of distributors not wanting
>to stock 17 versions of the same program.

Please tell that to the people in comp.os.linux.....
I give up.

>-Rick
>--
>Rick Richardson		Senior Staff Engineer	Writing X Widgets?
>DigiBoard, Inc.		Email: r...@digibd.com	Get Bert Bos's 'wbuild 2.0'
>6400 Flying Cloud Dr.	Fax:   (612) 943-0803	Any other way is primitive.
>Eden Prarie, MN 55344	Tel:   (612) 943-5383	<standard disclaimer>
-- 
  /--------------------------------------------------------------------------\
  | Mark A. Davis    | Lake Taylor Hospital | Norfolk, VA (804)-461-5001x431 |
  | Sys.Administrator|  Computer Services   | m...@taylor.wyvern.com   .uucp |
  \--------------------------------------------------------------------------/

Newsgroups: comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!sol.ctr.columbia.edu!math.ohio-state.edu!darwin.sura.net!
haven.umd.edu!uunet!noc.near.net!news.tufts.edu!gowen
From: go...@apex.cs.tufts.edu (Gregory Owen)
Subject: Re: Linux COFF (WAS: NT goes)
In-Reply-To: mark@taylor.uucp's message of Thu, 05 Aug 1993 20:41:48 GMT
Message-ID: <GOWEN.93Aug5212141@apex.cs.tufts.edu>
Sender: n...@news.tufts.edu (USENET News System)
Organization: Tufts University Department of Computer Science
References: <1993Aug02.180439.11053@taylor.uucp> <mebCB6E6u.JB2@netcom.com>
	<1993Aug03.143825.113@taylor.uucp> <CB8nz8.1D0@flatlin.ka.sub.org>
	<rick.744526681@digibd> <1993Aug05.204148.7419@taylor.uucp>
Date: Fri, 6 Aug 1993 02:21:41 GMT
Lines: 10


	In the announcement for the new SLS Linux release, I note that
ELF executable capablility is listed as a new feature.  It sounds more
like a package rather than a default, but as noted earlier, is
probably a step in the standardization direction for Linux.  I'd love
to hear more about it from anyone who has got it, but probably that
would be more appreciated on comp.os.linux.*.

  Greg Owen  { go...@forte.cs.tufts.edu, go...@xis.xerox.com }
1.01 GCS/GO d++ p+ c++ l++ u++ e+ -m+ s++/- n- h !(f)? g+ -w+ t+ r-- y?

Xref: gmd.de comp.unix.misc:6657 comp.unix.sys5.r4:4572 comp.os.linux:54852
Newsgroups: comp.unix.misc,comp.unix.sys5.r4,comp.os.linux
Path: gmd.de!Germany.EU.net!mcsun!sunic!uunet!europa.eng.gtefsd.com!
darwin.sura.net!math.ohio-state.edu!cs.utexas.edu!utnut!utzoo!telly!evan
From: ev...@telly.on.ca (Evan Leibovitch)
Subject: Is this becoming comp.linux.advocacy?
Message-ID: <CBCGxz.4I8@telly.on.ca>
Date: Fri, 6 Aug 1993 15:41:46 GMT
References: <WAYNE.93Aug1202432@backbone.uucp> <CB4oto.3F9@gator.oau.org> 
<1993Aug2.204307.26565@datamark.co.nz>
Organization: Somewhere just far enough out of Toronto
Lines: 102

My interest in Linux has not declined. Howver, my respect for those who
publicly exhibit massive ignorance of the marketplace, in the evangelical
drive to cheer Linux and knock everything else, has taken a nosedive.

While I believe Linux is a fantastic project, and it certainly has its
place as a "Unix for the masses", those who have taken the opportunity
to bash the commercial atrenatives as obsolete are definitely talking
from a software-induced stupor.

This has clearly become an advocacy issue. Please take this thread to the
Linux group, where its weenies can argue amongst themselves how Linux is
going to drive everything else into the ground. The rest of us have lives.

In article <1993Aug2.2...@datamark.co.nz>
	tho...@datamark.co.nz (Thomas Beagle) writes:

>In article <CB4ot...@gator.oau.org>
	la...@gator.oau.org (Larry D Snyder) writes:

>>how about drivers?  Are digiboard drivers available for linux?
>>How about the selection of hardware add-on products which exist
>>for other variations of Unix?  I like Word Perfect for writing,
>>how about WP for Linux -- fat chance.

>Well, consider the crappy device support that Unixware has. This is a
>commercial release.

Yes, let's consider them. What's wrong with them? What's wrong with the
device support of SCO, Wanna compare SCO's compatibility list with
Linux's?

What's a bigger issue is which OSs are attracting the support of third
parties. The drivers for Racal and Digiboard and Corollory and much other
hardware comes from the hardware maker, not the OS. How many of them are
writing Linux drivers?

>I've got the application server install pack sitting here but I can't
>install because it only supports Adaptec and DPT SCSI controllers.

Your analysis of Univel that brought you to the conclusion that only
Adaptec and DPT was supported is clearly wrong, and suggests to me you
made no effort to really check out the compatibility of the OS. How can
your arguments be considered credible when you so bias your assumptions?

Of the companies making host adaptors, Buslogic, DPT, Futute Domain,
Ultrastor and others supply their own Unix drivers. How many of these
companies supply Linux drivers?

Of course, it might make sense that a toy oprating system will have
better support for toy peripherals like soundblasters and el-cheapo
host adaptors. A cheap shot, sure, but totally appropriate as a reply to
the list of cheap shots that started this.

It's clear that the Linux OS will have to offer its own drivers for any
peripherals it wants to support, 'cause on the whole they're not going
to come from the hardware makers.

>And their video card support is a lot worse than that of Linux. (Which
>piggybacks on XFree86 (or whatever it's called)).

Despite a deliberate ignorance of the state of the world (XFree86 not
only runs on R4 systems, it's being developed on them, with Linux as an
afterthought), this ignores the issue of choice: what's available.

X server choices on Linux: XFree86

X server choices on R4: Stock servers from the OS vendor, XFree86, SCGS,
Pittsburgh Powercomputing, MetroLink, maybe others.

Try to run an ATI Graphics Ultra on Linux right now and see how far you get.
XFree86 may support it in a future release; commercial vendors like
MetroLink support it *now* on Sys5r4. These companies will likely always be
a few steps ahead of the XFree86 team on new hardware, because it's those
companies' bread and butter. Literally.

Commercial vendors looking at the market will develop for those
environments in which they can make a buck. It's reasonable for them to
assume that people who want their Unix for free aren't likely to spend
even $100 on an X server for accelerated cards -- so even if Linux is on
their porting list, it's on the bottom.

No matter how big the Linux installed base gets, its users will be
perceived as those who aren't interested in paying for software, and
resist software supplied without source. So commercial vendors will
have little interest in these users, no matter what their numbers.

For someone whose interest in Unix doesn't include commercial software,
Linux is a fantasic environment. Use sc as a spreadsheet,  TeX or BSD
troff along with your favourite editor for word processing, University
Ingres or perl as your database, and you can have a fairly good system
upon which a lot of useful work can be done. And you'll still have
source for everything....

But don't go spreading the manure that suggests Linux will ever be taken
seriously by any significant number of commercial application or driver
vendors. There's absolutely no basis for anyone to believe this except
evangelical fervor.

-- 
 Evan Leibovitch, Sound Software Ltd., located in beautiful Brampton, Ontario
         ev...@telly.on.ca / uunet!utzoo!telly!evan / (416) 452-0504
     "It costs a lot of money to look as cheap as I do" -- Dolly Parton

Xref: gmd.de comp.unix.misc:6681 comp.unix.sys5.r4:4600 comp.os.linux:54996
Newsgroups: comp.unix.misc,comp.unix.sys5.r4,comp.os.linux
Path: gmd.de!rrz.uni-koeln.de!news-rhrz!mpifr-bonn.mpg.de!uniol!
math.fu-berlin.de!news.dfn.de!scsing.switch.ch!univ-lyon1.fr!zaphod.crihan.fr!
pipex!uunet!news.gtech.com!node_man!ayr
From: ayr@node_man.uucp (Aleksey Y. Romanov)
Subject: Ho much USL license(was Re: Is this becoming comp.linux.advocacy)
Message-ID: <1993Aug7.160721.27689@gtech.com>
Sender: ne...@gtech.com (USENET Administrator)
Organization: GTECH Corporation, West Greenwich, RI
References: <CB4oto.3F9@gator.oau.org> <1993Aug2.204307.26565@datamark.co.nz> 
<CBCGxz.4I8@telly.on.ca>
Date: Sat, 7 Aug 1993 16:07:21 GMT
Lines: 81

In article <CBCGx...@telly.on.ca> ev...@telly.on.ca (Evan Leibovitch) writes:
>
>Commercial vendors looking at the market will develop for those
>environments in which they can make a buck. It's reasonable for them to
>assume that people who want their Unix for free aren't likely to spend
>even $100 on an X server for accelerated cards -- so even if Linux is on
>their porting list, it's on the bottom.
>
>No matter how big the Linux installed base gets, its users will be
>perceived as those who aren't interested in paying for software, and
>resist software supplied without source. So commercial vendors will
>have little interest in these users, no matter what their numbers.
>
>For someone whose interest in Unix doesn't include commercial software,
>Linux is a fantasic environment. Use sc as a spreadsheet,  TeX or BSD
>troff along with your favourite editor for word processing, University
>Ingres or perl as your database, and you can have a fairly good system
>upon which a lot of useful work can be done. And you'll still have
>source for everything....
>
>But don't go spreading the manure that suggests Linux will ever be taken
>seriously by any significant number of commercial application or driver
>vendors. There's absolutely no basis for anyone to believe this except
>evangelical fervor.

Please do not flame too hard if I am wrong :).

I agree with Evan. But .... Linux has a big advantage over UNIX in the 
area of development resources. There are thousands of Linux and GNU
enthusiasts and there are tens of engineers developing commercial
UNIX.  And among these engineers there are not many whose business is
bug fixing. So, as a result We have an advanced standard UNIX environment
which is too buggy from my point of view. Can the advantages of
Linux and commercial UNIX be unified in a single product ?

How much is USL(or Univel) source license ? I heard that it is below
$200,000.  I guess that it is possible to find 200 unix-fans (each will
contribute $1000) in order to found a UNIX company.

The charter of this company has to include:

1. To fix bugs in kernel and utilities and to sell fixed drop-in
   UNIX component replacements on the commercial market,
   The emphasis shall be made on kernel bug fixing and
   drop-in replacements from PD (of free) software.
   The latter one will insure the wider use of free software
   componenets in order to increase development resource base:
   the guy whose job is to port network utilities from say NET2
   release does not have to have access to the licensed source
   code from USL.

2. The Sane UNIX Project. The goal of this project is to make
   an offering of low cost (say below $200 per node)
   PC networking environment. By combining
   in the single product the best of: the debugged core of commercial
   UNIX system, free software, commercial software from different
   vendors (example Motif from OSF, not Moolit). No decision
   has to be made due to NIH decease.


I do expect this to have a form of spare time activity for contributors
at least for a first time.

May be Solaris is a better candidate for such a project I do not know.

Once again, please, do not flame me too much if am totally wrong.


> Evan Leibovitch, Sound Software Ltd., located in beautiful Brampton, Ontario
>         ev...@telly.on.ca / uunet!utzoo!telly!evan / (416) 452-0504
>     "It costs a lot of money to look as cheap as I do" -- Dolly Parton

Aleksey



-- 
Aleksey Y. Romanov

DISCLAIMER: The views expressed are entirely my own and do not
necessary reflect those of my employer or anyone else

Xref: gmd.de comp.unix.misc:6688 comp.unix.sys5.r4:4606 comp.os.linux:55012
Newsgroups: comp.unix.misc,comp.unix.sys5.r4,comp.os.linux
Path: gmd.de!xlink.net!math.fu-berlin.de!ifmsun8.ifm.uni-hamburg.de!
rzsun2.informatik.uni-hamburg.de!news.dfn.de!darwin.sura.net!math.ohio-state.edu!
uwm.edu!spool.mu.edu!agate!doc.ic.ac.uk!pipex!uunet!world!entropy
From: ent...@world.std.com (Lawrence Foard)
Subject: Re: Ho much USL license(was Re: Is this becoming comp.linux.advocacy)
Message-ID: <CBEtu2.II4@world.std.com>
Organization: The World Public Access UNIX, Brookline, MA
References: <CB4oto.3F9@gator.oau.org> <1993Aug2.204307.26565@datamark.co.nz> 
<CBCGxz.4I8@telly.on.ca> <1993Aug7.160721.27689@gtech.com>
Date: Sat, 7 Aug 1993 22:15:38 GMT
Lines: 55

In article <1993Aug7.1...@gtech.com> ayr@node_man.uucp (Aleksey Y. Romanov) writes:
>
>I agree with Evan. But .... Linux has a big advantage over UNIX in the 
>area of development resources. There are thousands of Linux and GNU
>enthusiasts and there are tens of engineers developing commercial
>UNIX.  And among these engineers there are not many whose business is
>bug fixing. So, as a result We have an advanced standard UNIX environment
>which is too buggy from my point of view. Can the advantages of
>Linux and commercial UNIX be unified in a single product ?
>
>How much is USL(or Univel) source license ? I heard that it is below
>$200,000.  I guess that it is possible to find 200 unix-fans (each will
>contribute $1000) in order to found a UNIX company.

Why bother when Linux is already free and better than commercial Unix? I'm
not going to waste time pandering to peoples biases against free software,
Linux is the best development environment that I have ever used. And the
most reasonably licensed of any product available.  

>
>The charter of this company has to include:
>
>1. To fix bugs in kernel and utilities and to sell fixed drop-in
>   UNIX component replacements on the commercial market,
>   The emphasis shall be made on kernel bug fixing and
>   drop-in replacements from PD (of free) software.
>   The latter one will insure the wider use of free software
>   componenets in order to increase development resource base:
>   the guy whose job is to port network utilities from say NET2
>   release does not have to have access to the licensed source
>   code from USL.
>
>2. The Sane UNIX Project. The goal of this project is to make
>   an offering of low cost (say below $200 per node)
>   PC networking environment. By combining
>   in the single product the best of: the debugged core of commercial
>   UNIX system, free software, commercial software from different
>   vendors (example Motif from OSF, not Moolit). No decision
>   has to be made due to NIH decease.
>
>
>I do expect this to have a form of spare time activity for contributors
>at least for a first time.
>
>May be Solaris is a better candidate for such a project I do not know.
>
>Once again, please, do not flame me too much if am totally wrong.

I fail to see the purpose? What advantage does this $200 Unix have
over Linux? 
-- 
------ Time: 820713600 seconds, Space: 1727.2mm, Mass: 9.5E10 ug         . 
\    / Kinsey: 4.5, Religion: Science, Energy: 8.55E18, OS: Linux       . .     
 \  / Species: Human, Planet: Earth, Fame: Tinymush, Allergy: Dustmite . . .     
  \/ Purity: 40, Sex: male, frequent, Drugs: Caffeine, Rock & roll:   . . . .     

Xref: gmd.de comp.unix.misc:6691 comp.unix.sys5.r4:4609 comp.os.linux:55029
Newsgroups: comp.unix.misc,comp.unix.sys5.r4,comp.os.linux
Path: gmd.de!xlink.net!sol.ctr.columbia.edu!math.ohio-state.edu!darwin.sura.net!
haven.umd.edu!uunet!tarpit!gator!not-for-mail
From: la...@gator.oau.org (Larry D Snyder)
Subject: Re: Ho much USL license(was Re: Is this becoming comp.linux.advocacy)
Message-ID: <CBF3pn.GFE@gator.oau.org>
Organization: Gator Communications, Orlando, Florida
References: <CB4oto.3F9@gator.oau.org> <1993Aug2.204307.26565@datamark.co.nz> 
<CBCGxz.4I8@telly.on.ca> <1993Aug7.160721.27689@gtech.com> 
<CBEtu2.II4@world.std.com>
Date: Sun, 8 Aug 1993 01:48:58 GMT
Lines: 20

ent...@world.std.com (Lawrence Foard) writes:

>Why bother when Linux is already free and better than commercial Unix? I'm

That's a personal opinion.  Linux lacks applications and support
from third party vendors.   Try finding a smart multiport board from
Digiboard with drivers for Linux.  Digiboard has never heard of Linux
according to their technical support.  Call Word Perfect and order WP
for Linux.  Call Ingress and ask for their database package for Linux.


>I fail to see the purpose? What advantage does this $200 Unix have
>over Linux? 

End user support, applications and support from industry known third party
hardware and software Unix vendors.  

-- 
Larry Snyder                                    Internet: la...@gator.oau.org
Orlando, Florida                            UUCP: ..!uunet!tarpit!gator!larry

Newsgroups: comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!howland.reston.ans.net!vixen.cso.uiuc.edu!uchinews!
cs.umd.edu!ra!tantalus.nrl.navy.mil!eric
From: e...@tantalus.nrl.navy.mil (Eric Youngdale)
Subject: Re: Linux COFF (WAS: NT goes)
Message-ID: <CBKJEM.4B2@ra.nrl.navy.mil>
Sender: use...@ra.nrl.navy.mil
Organization: Naval Research Laboratory
References: <rick.744526681@digibd> <1993Aug05.204148.7419@taylor.uucp> 
<GOWEN.93Aug5212141@apex.cs.tufts.edu>
Date: Wed, 11 Aug 1993 00:15:58 GMT
Lines: 94

In article <GOWEN.93Aug5212...@apex.cs.tufts.edu> go...@apex.cs.tufts.edu 
(Gregory Owen) writes:
>
>	In the announcement for the new SLS Linux release, I note that
>ELF executable capablility is listed as a new feature.  It sounds more
>like a package rather than a default, but as noted earlier, is
>probably a step in the standardization direction for Linux.  I'd love
>to hear more about it from anyone who has got it, but probably that
>would be more appreciated on comp.os.linux.*.

	As I am the author of this package, I guess I am in the best position
to comment.  It is basically an add-on package.  The linux kernel can recognize
ELF binaries and mmap them to the proper address, and it can also try and load
the program interpreter that is requested by the ELF binary (usually
/usr/lib/libc.so.1).  A program by this name is provided that does the dynamic
linking required so that the image will run, and then control is passed to the
application.  Currently there are a couple of libraries that the dynamic linker
resolves the externals to, although ultimately most calls go through the
regular linux libraries.  I should point out that the program loader is capable
of loading ELF shared libraries as well.

	I can run a number of small to medium size SVr4 binaries under linux,
the largest one that I have bothered to try is the GNU assembler.  There are,
of course a number of limitations and problems that remain (and I know that it
is relatively easy to find binaries that will not work because of incomplete
emulation), but in the long term my thinking is that a separate version of the
linux libc will be required to run ELF binaries (generally errno handling
requires this, since the numbering is a little bit different), and in general
things will be much cleaner once this happens.  I expect that there will only
be support for things for which there is an analagous feature in the linux
kernel - STREAMS will probably not be supported for example (as far as I know,
no one has yet expressed a serious interest in getting STREAMS working under
Linux).

	People have speculated that there is a long term goal for linux to the
ELF format switch from the a.out format that is currently in use.  I would like
to confirm that there has been some discussion in this regard, and I personally
would like to see it go in this direction.  Even if this happens, backwards
compatibility will continue to be provided in order to run the older a.out
binaries.  I do not expect that much kernel bloat will result from the
capability of running several different binary formats - the kernel is now
quite modular so that supporting a new binary format simply means dropping in a
new function to recognize and load a file in the binary format.  The main thing
that is needed now for ELF is for the GNU binutils to be able to reliably deal
with ELF format files and generate shared libraries - I expect that this is
simply a matter of time.

	As far as commercial shrink-wrapped applications are concerned, I do
indeed recognize that supporting COFF in some form would be more logical than
merely supporting ELF.  The linux kernel now has a "lcall 7,0" entry point,
although it does not do very much right now (linux normally uses an "int 0x80"
for syscalls, so there is no conflict between the different syscall conventions
used by iBCS2 and linux).  There are patches to the linux kernel to handle
some of the simple syscalls (for which there are directly compatible analogs,
or for which the conversion of data formats is trivial), and work is now
getting underway to complete this.  I doubt that all of the iBSC2/SCO featurism
will be provided, however one of the most common requests is to be able to run
something like WordPerfect, so I would expect that the syscall functionality
required to run such popular applications would be the first priority.

	Finally, a number of SVr4 people have criticized linux by saying that
the frequency of updates makes commercial software development impractical.  In
response to this I should point out that while there are a lot of updates, one
of the upmost concerns for the developers is to maintain backward
compatibility.  Even for free software, it is an incredible pain in the neck to
have to replace binaries all over the system, so it would be very
counterproductive to break a bunch of binaries just to add some new feature of
some kind.  Even when backward compatibility is not possible, the worst case is
almost always that you merely have to recompile and relink.  It is of course
possible that new features will make cleaner or better implementations of some
applications possible, but there would be no reason to recode something unless
you really wanted to.

	I should add in passing that I do use SVr4 at work.  When we bought our
machines, we got our SVr4 from Dell.  Every so often, I run across a bug of
some kind or another, and while these are usually not show-stoppers, they are
annoying.  It is not at all clear that bugfixes even exist for these problems
(and usually these are the sort of bugs that I could fix in about 30 minutes if
I had the source code).  Getting help from Dell is close to impossible now -
(just getting a human to answer the phone or call you back is quite a feat in
itself).  Thus I end up writing some stupid hack to work around the bug and end
up spending far more time in the process than I would have if I had the source
handy.  Dell is not the only one who I could complain about - we have a Fortran
compiler which we got from a 3rd party vendor for which we payed good money,
and it crashes (i.e. compiler dumps core) when given input code in a certain
syntax.  As far as I can tell the vendor is doing no development work or even
providing bugfixes.  Instead I have to tell people here to "rearrange" their
fortran source code to work around the compiler crash.  One guy got so fed up
that he started using f2c for production code.  Bleah.  This is supposed to be
better??????

-Eric
-- 
"When Gregor Samsa woke up one morning from unsettling dreams, he
found himself changed in his bed into a lawyer."

Newsgroups: comp.unix.misc,comp.unix.sys5.r4
Path: gmd.de!xlink.net!howland.reston.ans.net!math.ohio-state.edu!usc!
elroy.jpl.nasa.gov!decwrl!uunet!wyvern!taylor.wyvern.com!mark
From: m...@taylor.uucp (Mark A. Davis)
Subject: Re: Linux COFF (WAS: NT goes)
Organization: Lake Taylor Hospital Computer Services
Date: Wed, 11 Aug 1993 14:48:50 GMT
Message-ID: <1993Aug11.144850.17180@taylor.uucp>
References: <rick.744526681@digibd> <1993Aug05.204148.7419@taylor.uucp> 
<GOWEN.93Aug5212141@apex.cs.tufts.edu> <CBKJEM.4B2@ra.nrl.navy.mil>
Lines: 40

e...@tantalus.nrl.navy.mil (Eric Youngdale) writes:

>	As I am the author of this package, I guess I am in the best position
>to comment. 

[...]

>	People have speculated that there is a long term goal for linux to the
>ELF format switch from the a.out format that is currently in use.  I would like
>to confirm that there has been some discussion in this regard, and I personally
>would like to see it go in this direction.

You get suport from me.  I think ELF native would be an EXCELLENT idea.  It
should also reduce the needs for separate "compatibility libraries" and such.
It might also make COFF easier to work with?

>	As far as commercial shrink-wrapped applications are concerned, I do
>indeed recognize that supporting COFF in some form would be more logical than
>merely supporting ELF.  The linux kernel now has a "lcall 7,0" entry point,
>although it does not do very much right now (linux normally uses an "int 0x80"
>for syscalls, so there is no conflict between the different syscall conventions
>used by iBCS2 and linux).  There are patches to the linux kernel to handle
>some of the simple syscalls (for which there are directly compatible analogs,
>or for which the conversion of data formats is trivial), and work is now
>getting underway to complete this.  I doubt that all of the iBSC2/SCO featurism
>will be provided, however one of the most common requests is to be able to run
>something like WordPerfect, so I would expect that the syscall functionality
>required to run such popular applications would be the first priority.

Hot Damn!  As I am sure you already know by now from all my blabbering in
c.o.l I think COFF should be a high priority.  Do you think the problems
it faces will be overcome?  I'm mostly worried about 3.2 X lib's too, seems
like running COFF with X programs could get really sticky.

Get ready for my mental support beam..... bzzzzzzzzzzzzzzzzzzzzzzzzzzz
-- 
  /--------------------------------------------------------------------------\
  | Mark A. Davis    | Lake Taylor Hospital | Norfolk, VA (804)-461-5001x431 |
  | Sys.Administrator|  Computer Services   | m...@taylor.wyvern.com   .uucp |
  \--------------------------------------------------------------------------/

Xref: gmd.de comp.unix.misc:6780 comp.unix.sys5.r4:4728 comp.os.linux:55860
Path: gmd.de!xlink.net!sol.ctr.columbia.edu!math.ohio-state.edu!cs.utexas.edu!
uunet!pipex!sunic!news.funet.fi!funic!nntp.hut.fi!cs.hut.fi!arl
From: a...@cs.hut.fi (Ari Lemmke)
Newsgroups: comp.unix.misc,comp.unix.sys5.r4,comp.os.linux
Subject: Re: Ho much USL license(was Re: Is this becoming comp.linux.advocacy)
Date: 12 Aug 1993 10:59:46 GMT
Organization: Helsinki University of Technology, Finland
Lines: 18
Distribution: inet
Message-ID: <ARL.93Aug12140128@deathstar.cs.hut.fi>
References: <CB4oto.3F9@gator.oau.org> <1993Aug2.204307.26565@datamark.co.nz>
	<CBCGxz.4I8@telly.on.ca> <1993Aug7.160721.27689@gtech.com>
	<CBEtu2.II4@world.std.com> <CBF3pn.GFE@gator.oau.org>
NNTP-Posting-Host: deathstar.cs.hut.fi
In-reply-to: larry@gator.oau.org's message of Sun, 8 Aug 1993 01:48:58 GMT

In article <CBF3p...@gator.oau.org> la...@gator.oau.org (Larry D Snyder) writes:
:   That's a personal opinion.  Linux lacks applications and support
:   from third party vendors.   Try finding a smart multiport board from
:   Digiboard with drivers for Linux.  Digiboard has never heard of Linux
:   according to their technical support.  Call Word Perfect and order WP
:   for Linux.  Call Ingress and ask for their database package for Linux.

	It will soon change, just create the demand.

:   End user support, applications and support from industry known third party
:   hardware and software Unix vendors.  

	And lot's of work you do not own.

: Larry Snyder                                    Internet: la...@gator.oau.org

	arl