[TUHS] System Economics (was is Linux "officially branded UNIX")

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1328 bytes --]

On Tue, Mar 14, 2017 at 7:35 AM, Tim Bradshaw <tfb at tfeb.org> wrote:

> But the people who have spent 9-figure sums on all this
> marginally-functional tin that the Unix vendors foisted on them don't
> look at it that way: they just want something which is not Unix, and
> which runs on cheap tin.
>
​Fair enough -- but I think that this is really another way of describing
Prof. Christiansen's disruption theory​.   The "lessor" technology wins
over "better" technology because it's good enough.

I'm curious for the Banks, in your experience - which were the UNIX vendors
that were pushing 9-figure UNIX boxes.  I'll guess, IBM was one of them.
Maybe NCR.     What HP, Sun, DEC in that bundle?



>  Linux is not Unix, and runs on cheap tin.
>
I
​believe that
 the point you are making is that "white box" PC's running a UNIX-like
system - aka Linux could comes pretty close to doing what the highly touted
AIX, NCR et al were doing and were "good enough" to get the job done.

​And that's not a statement about UNIX as much as a statement about, the
WINTEL ecosystem, that Linux sat on top of and did an extremely impressive
job of utilizing.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170314/fe10d73b/attachment-0001.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Larry McVoy]

On Tue, Mar 14, 2017 at 10:43:51AM -0400, Clem Cole wrote:
> On Tue, Mar 14, 2017 at 7:35 AM, Tim Bradshaw <tfb at tfeb.org> wrote:
> >  Linux is not Unix, and runs on cheap tin.
> >
> I ???believe that
>  the point you are making is that "white box" PC's running a UNIX-like
> system - aka Linux could comes pretty close to doing what the highly touted
> AIX, NCR et al were doing and were "good enough" to get the job done.

As someone who dedicated a bunch of his life to Unix, it pains me to say
it but Linux is better than a lot of the Unix systems from back in the
day.  I loved SunOS but I wouldn't trade today's Linux for SunOS and I
don't think there are very many people who would disagree.

It got better than "good enough".

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Arthur Krewat]

But how far along are we on the Linux timeline, and how far along was 
Sun on the SunOS timeline before they stopped developing it?

It's been 23 or so years since my first exposure to Linux.

SunOS started at 1.0 in 1983, and last release was just before 1995. 12 
years in total. From https://en.wikipedia.org/wiki/SunOS

Now, of course, I understand SunOS is based on BSD so there is a lot 
more work invested in SunOS before Sun even started on it which adds 
another 10 years (maybe less) to the SunOS development timeline. But in 
reality, how much of Linux was based on previous works?

Just a thought experiment, nothing more.


On 3/14/2017 11:38 AM, Larry McVoy wrote:
> As someone who dedicated a bunch of his life to Unix, it pains me to say
> it but Linux is better than a lot of the Unix systems from back in the
> day.  I loved SunOS but I wouldn't trade today's Linux for SunOS and I
> don't think there are very many people who would disagree.
>
> It got better than "good enough".
>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Larry McVoy]

SunOS wasn't multi threaded.  Linux seems to have done that pretty well
without getting all bloated (unlike early Solaris releases, I can't
speak to the later ones). 

Linux is just more mature, has had more people working on it (which is
both a good and a bad thing).  And didn't have Sun's stick in the mud
approach to compat that made things like /proc in Solaris way way way
less useful than Linux' /proc.

So it's really hard to say.

On Tue, Mar 14, 2017 at 11:51:47AM -0400, Arthur Krewat wrote:
> But how far along are we on the Linux timeline, and how far along was Sun on
> the SunOS timeline before they stopped developing it?
> 
> It's been 23 or so years since my first exposure to Linux.
> 
> SunOS started at 1.0 in 1983, and last release was just before 1995. 12
> years in total. From https://en.wikipedia.org/wiki/SunOS
> 
> Now, of course, I understand SunOS is based on BSD so there is a lot more
> work invested in SunOS before Sun even started on it which adds another 10
> years (maybe less) to the SunOS development timeline. But in reality, how
> much of Linux was based on previous works?
> 
> Just a thought experiment, nothing more.
> 
> 
> On 3/14/2017 11:38 AM, Larry McVoy wrote:
> >As someone who dedicated a bunch of his life to Unix, it pains me to say
> >it but Linux is better than a lot of the Unix systems from back in the
> >day.  I loved SunOS but I wouldn't trade today's Linux for SunOS and I
> >don't think there are very many people who would disagree.
> >
> >It got better than "good enough".
> >

-- 
---
Larry McVoy            	     lm at mcvoy.com             http://www.mcvoy.com/lm 

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Michael Kjörling]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 478 bytes --]

On 14 Mar 2017 11:51 -0400, from krewat at kilonet.net (Arthur Krewat):
> in reality, how much of Linux was based on previous works?

Linux the kernel, or Linux the usable operating system (which would
include at least the essential userspace parts)?

-- 
Michael Kjörling • https://michael.kjorling.se • michael at kjorling.se
                 “People who think they know everything really annoy
                 those of us who know we don’t.” (Bjarne Stroustrup)

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Arthur Krewat]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 544 bytes --]

Everything. I'm trying to grok how long Linux as a whole was in active 
development. That includes all the GNU utilities, GCC, everything.

Just like a "regular" corporate development environment would have 
devoted to the cause :)



On 3/14/2017 11:57 AM, Michael Kjörling wrote:
> On 14 Mar 2017 11:51 -0400, from krewat at kilonet.net (Arthur Krewat):
>> in reality, how much of Linux was based on previous works?
> Linux the kernel, or Linux the usable operating system (which would
> include at least the essential userspace parts)?
>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Jason Stevens]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1983 bytes --]

GCC started in 1986 with the 0.9 release in 1988, along with gas, and binutils.  It originally targeted the 68020 and the VAX.  Naturally more platforms were added.

GCC 0.9 can be found, but many of the early versions until 1.21 seem to have been lost.  I've been on and off trying to catalog some of this stuff, as you pointed out there was a lot of ground work getting Minix on the 8086, then the Bruce Evans 80386 port of Minix which then could be used to cross compile Linux using GCC 1.40...  Although using the DJGPP MS-DOS port of GCC as a template I am able to build early Linux kernels on Windows using the old FSF GCC and binutils.  So theoretically it could be cross compiled from MS-DOS.

And there is of course, the original Libc, and bash which was the original environment, then later the GNU filesystem utils.

oldinux.org has many of these old software artifacts to check out, along with vim.org

http://ftp.vim.org/languages/gcc/old-releases/gcc-1/

And here for some binutils going back to 1988

https://www.mirrorservice.org/sites/sourceware.org/pub/binutils/old-releases/

On March 15, 2017 12:20:06 AM GMT+08:00, Arthur Krewat <krewat at kilonet.net> wrote:
>Everything. I'm trying to grok how long Linux as a whole was in active 
>development. That includes all the GNU utilities, GCC, everything.
>
>Just like a "regular" corporate development environment would have 
>devoted to the cause :)
>
>
>
>On 3/14/2017 11:57 AM, Michael Kjörling wrote:
>> On 14 Mar 2017 11:51 -0400, from krewat at kilonet.net (Arthur Krewat):
>>> in reality, how much of Linux was based on previous works?
>> Linux the kernel, or Linux the usable operating system (which would
>> include at least the essential userspace parts)?
>>

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/8be7a0e6/attachment-0001.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 931 bytes --]

On Tue, Mar 14, 2017 at 2:06 PM, Jason Stevens <
jsteve at superglobalmegacorp.com> wrote:

> ​i​
> t originally targeted the 68020


​It ran on the 68000 before the *20.   rms had access to Masscomp box we
gave him fairly early on.​ I remember pitching to our exec's trying to get
him more HW and trying explain who he was to them at the time.  That would
have been late 85.     He may have had access to that system before he got
a Sun but I don't remember.  That said, I'm sure the MC-500 was not the
first 68000 he had access.  I think he was using HW in Steve Ward's lab
that the Trix guys were developing with TI and he might have had access to
an Apollo system.  If we can find Jack test he might remember,  Noel do you
remember how that went down?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170314/dccdd74e/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Jason Stevens]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1396 bytes --]

It would certainly explain the -m68000 flags, in the announcement it only mentioned the sun and Vax

https://groups.google.com/forum/m/#!topic/mod.compilers/ynAVuwR7dPw

As always the more info from prior to the 0.9 announcement is interesting!

On March 15, 2017 2:31:32 AM GMT+08:00, Clem Cole <clemc at ccc.com> wrote:
>
>On Tue, Mar 14, 2017 at 2:06 PM, Jason Stevens <
>jsteve at superglobalmegacorp.com <mailto:jsteve at superglobalmegacorp.com>
>>
>wrote:
>
>
>​i​
>t originally targeted the 68020
>
>
>​It ran on the 68000 before the *20.   rms had access to Masscomp box
>we
>gave him fairly early on.​ I remember pitching to our exec's trying to
>get him more HW and trying explain who he was to them at the time. 
>That
>would have been late 85.     He may have had access to that system
>before he got a Sun but I don't remember.  That said, I'm sure the
>MC-500 was not the first 68000 he had access.  I think he was using HW
>in Steve Ward's lab that the Trix guys were developing with TI and he
>might have had access to an Apollo system.  If we can find Jack test he
>might remember,  Noel do you remember how that went down?

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/a122ff90/attachment-0001.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1016 bytes --]

That's not really fair.   The compiler and the utilities was developed for
UNIX long before the Linux kernel existed.  This is why so many of consider
"Linux" just the current version of UNIX.

On Tue, Mar 14, 2017 at 12:20 PM, Arthur Krewat <krewat at kilonet.net> wrote:

> Everything. I'm trying to grok how long Linux as a whole was in active
> development. That includes all the GNU utilities, GCC, everything.
>
> Just like a "regular" corporate development environment would have devoted
> to the cause :)
>
>
>
>
> On 3/14/2017 11:57 AM, Michael Kjörling wrote:
>
>> On 14 Mar 2017 11:51 -0400, from krewat at kilonet.net (Arthur Krewat):
>>
>>> in reality, how much of Linux was based on previous works?
>>>
>> Linux the kernel, or Linux the usable operating system (which would
>> include at least the essential userspace parts)?
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170314/5a374d7f/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Arthur Krewat]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2305 bytes --]

So what I'm hearing is Linux's timeline, which includes things that were 
not developed just for Linux, extends further out than SunOS does.

Again, I'm including everything ... You could make a case for certain 
Unixes that do not include a pre-existing C compiler being bounded by 
their own development (or any other operating system that needs a 
precursor).  For example, say there was an operating system that used a 
C compiler to build itself that was developed 10 years before. That 
example operating system's timeline would have to include said C 
compiler IMHO.

On the other hand, an operating system who's sole method of creation was 
engineered in year 0, and was "developed" for 10 years and ended, we 
could say that OS's timeline was a solid 10 years.

All I'm saying is comparing Linux's timeline to something like SunOS has 
to include everything that went into both because they both relied on 
precursors.

Side note: I'm a bit of a bitch when it comes to Linux - which doesn't 
mean I don't think Linux is "UNIX" - it just means I think it's the 
Coherent of today's UNIX ;)


On 3/14/2017 2:20 PM, Clem Cole wrote:
> That's not really fair.   The compiler and the utilities was developed 
> for UNIX long before the Linux kernel existed.  This is why so many of 
> consider "Linux" just the current version of UNIX.
>
> On Tue, Mar 14, 2017 at 12:20 PM, Arthur Krewat <krewat at kilonet.net 
> <mailto:krewat at kilonet.net>> wrote:
>
>     Everything. I'm trying to grok how long Linux as a whole was in
>     active development. That includes all the GNU utilities, GCC,
>     everything.
>
>     Just like a "regular" corporate development environment would have
>     devoted to the cause :)
>
>
>
>
>     On 3/14/2017 11:57 AM, Michael Kjörling wrote:
>
>         On 14 Mar 2017 11:51 -0400, from krewat at kilonet.net
>         <mailto:krewat at kilonet.net> (Arthur Krewat):
>
>             in reality, how much of Linux was based on previous works?
>
>         Linux the kernel, or Linux the usable operating system (which
>         would
>         include at least the essential userspace parts)?
>
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170314/ef3c0c25/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Michael Kjörling]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2025 bytes --]

On 14 Mar 2017 15:48 -0400, from krewat at kilonet.net (Arthur Krewat):
> Again, I'm including everything ... You could make a case for
> certain Unixes that do not include a pre-existing C compiler being
> bounded by their own development (or any other operating system that
> needs a precursor).  For example, say there was an operating system
> that used a C compiler to build itself that was developed 10 years
> before. That example operating system's timeline would have to
> include said C compiler IMHO.
> 
> On the other hand, an operating system who's sole method of creation
> was engineered in year 0, and was "developed" for 10 years and
> ended, we could say that OS's timeline was a solid 10 years.

Then why limit yourself to the C compiler? The operating system
probably relies on an early bootstrapper layer to start (on the IBM PC
and similar systems this is the BIOS or more recently UEFI; other
architectures are similar or different). The code was probably written
using keyboards, which may or may not rely on firmware for the
physical key to key code to operating system input mapping, let alone
the editor and file system code used to store those first few chunks
of code. And what about the timelines of _those_? At some point the
system becomes self-hosting in the software sense, but it took work to
get to that point. And so on.

I think you see where I am heading with this; if we're going to
include things that were not done specifically for the operating
system in question, then unless we draw a clear line somewhere, we end
up with some guy working on vacuum tube theory a century ago and
_still_ aren't anywhere near an answer to "how long is the timeline of
this piece of software?". Hence, absent some kind of demarcation, that
discussion becomes meaningless.

-- 
Michael Kjörling • https://michael.kjorling.se • michael at kjorling.se
                 “People who think they know everything really annoy
                 those of us who know we don’t.” (Bjarne Stroustrup)

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Arthur Krewat]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2785 bytes --]

You make a valid point, and re-reading what I wrote, I find that I 
pushed the example too far :)

The subject was originally that SunOS at it's end-of-life did not have 
the features that Linux now does, and comparing their development 
lengths brings up an interesting question. What would SunOS have become 
if it had been actively developed for as long as Linux has? I was trying 
to make the point that SunOS didn't have the same amount of elapsed time 
invested in it's development, and yet in fairness it was based on BSD 
which adds to that elapsed time significantly.

Off-topic: Anyone ever run SunOS on a Sparc-10 or similar platform 
(670?) with two processors? Was it my imagination or did it actually use 
both processors?

Side note: I was one of those people who was pulled 
kicking-and-screaming into the Solaris (SVR4) world after having 
administered SunOS for years.


On 3/15/2017 10:32 AM, Michael Kjörling wrote:
> On 14 Mar 2017 15:48 -0400, from krewat at kilonet.net (Arthur Krewat):
>> Again, I'm including everything ... You could make a case for
>> certain Unixes that do not include a pre-existing C compiler being
>> bounded by their own development (or any other operating system that
>> needs a precursor).  For example, say there was an operating system
>> that used a C compiler to build itself that was developed 10 years
>> before. That example operating system's timeline would have to
>> include said C compiler IMHO.
>>
>> On the other hand, an operating system who's sole method of creation
>> was engineered in year 0, and was "developed" for 10 years and
>> ended, we could say that OS's timeline was a solid 10 years.
> Then why limit yourself to the C compiler? The operating system
> probably relies on an early bootstrapper layer to start (on the IBM PC
> and similar systems this is the BIOS or more recently UEFI; other
> architectures are similar or different). The code was probably written
> using keyboards, which may or may not rely on firmware for the
> physical key to key code to operating system input mapping, let alone
> the editor and file system code used to store those first few chunks
> of code. And what about the timelines of _those_? At some point the
> system becomes self-hosting in the software sense, but it took work to
> get to that point. And so on.
>
> I think you see where I am heading with this; if we're going to
> include things that were not done specifically for the operating
> system in question, then unless we draw a clear line somewhere, we end
> up with some guy working on vacuum tube theory a century ago and
> _still_ aren't anywhere near an answer to "how long is the timeline of
> this piece of software?". Hence, absent some kind of demarcation, that
> discussion becomes meaningless.
>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Arthur Krewat]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 684 bytes --]

Sorry, in this context, SunOS means 4.1.4 - not Solaris SVR4

I run Solaris myself, and love it.

On 3/15/2017 11:48 AM, Joerg Schilling wrote:
> Arthur Krewat <krewat at kilonet.net> wrote:
>
>> You make a valid point, and re-reading what I wrote, I find that I
>> pushed the example too far :)
>>
>> The subject was originally that SunOS at it's end-of-life did not have
>> the features that Linux now does, and comparing their development
>> lengths brings up an interesting question. What would SunOS have become
> So you believe that SunOS-5.11 is no longer alive?
>
> There is an Oracle based version and a OpenSolarisd based version developed by
> the community.
>
> Jörg
>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Larry McVoy]

On Wed, Mar 15, 2017 at 11:36:21AM -0400, Arthur Krewat wrote:
> Off-topic: Anyone ever run SunOS on a Sparc-10 or similar platform (670?)
> with two processors? Was it my imagination or did it actually use both
> processors?

Yeah, SunOS 4.1.4 had some MP work done to it.  Pretty sure I posted
about it here and dragged Greg Limes into it.  He was involved in that
work.  I think it sort of worked up to 4 CPUs but as with all early
kernel threading stuff it worked better when it was a 4 cpus of 
userland work, less so when it was 4 cpus of I/O.

> Side note: I was one of those people who was pulled kicking-and-screaming
> into the Solaris (SVR4) world 

You and me both.  

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Warner Losh]

On Wed, Mar 15, 2017 at 9:59 AM, Larry McVoy <lm at mcvoy.com> wrote:
> On Wed, Mar 15, 2017 at 11:36:21AM -0400, Arthur Krewat wrote:
>> Off-topic: Anyone ever run SunOS on a Sparc-10 or similar platform (670?)
>> with two processors? Was it my imagination or did it actually use both
>> processors?
>
> Yeah, SunOS 4.1.4 had some MP work done to it.  Pretty sure I posted
> about it here and dragged Greg Limes into it.  He was involved in that
> work.  I think it sort of worked up to 4 CPUs but as with all early
> kernel threading stuff it worked better when it was a 4 cpus of
> userland work, less so when it was 4 cpus of I/O.

Solbourne computer produced OS/MP, which was a SMPized version of
SunOS 4.1. It scaled to at least 16 CPUs. Userland was 100% compatible
with SunOS at the given revision level.  My team (the OI group) was
part of Solbourne for a while, so we got much of that gear when we
were spun out to ParcPlace. Our main build server had 12 CPUs, and it
was nice being able to do make -j 16. cfront was, for its time, quite
the pig. Now it's lightyears faster than clang, though produces lousy
code... OS/MP ran on hardware that was 50MHz SuperSparc CPUs and could
be configured up to 14 CPUs and a whopping 256MB of RAM...

Warner

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Larry McVoy]

On Wed, Mar 15, 2017 at 11:43:34AM -0600, Warner Losh wrote:
> Solbourne computer produced OS/MP, which was a SMPized version of
> SunOS 4.1. It scaled to at least 16 CPUs. 

I don't suppose that code is around anywhere?  I'd love to see what
they did.  Even as a set of diffs from 4.1 would be cool.

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Warner Losh]

On Wed, Mar 15, 2017 at 1:02 PM, Larry McVoy <lm at mcvoy.com> wrote:
> On Wed, Mar 15, 2017 at 11:43:34AM -0600, Warner Losh wrote:
>> Solbourne computer produced OS/MP, which was a SMPized version of
>> SunOS 4.1. It scaled to at least 16 CPUs.
>
> I don't suppose that code is around anywhere?  I'd love to see what
> they did.  Even as a set of diffs from 4.1 would be cool.

It was never released publicly. I know a guy who might still have a
copy, but I don't know if he'd be willing to let me copy it... There's
a small chance I still have a copy on a backup dump somewhere, but I
kinda doubt it...

Warner

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Warner Losh]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2263 bytes --]

Let's not forget X11 which has a long history as well starting in
1984. The 11th version of the protocol (X11) dates from 1987. All the
X11 versions are online still due to the X consortium. However, X10
and earlier can be hard to find. https://www.x.org/releases/ has X10R3
and X10R4, but nothing earlier. That's also a huge part of Linux since
it represents its windowing system. I used X10 on a sun 3/50 back in
the day before they upgraded it to X11. It was slower and buggier than
SunTools, but more cutting edge. suntools is dead and X11 is still
alive. suntools went directly to the frame buffer, while X always did
the protocol thing (though with many attempts over the years to make
the protocol layer optional, maybe wayland will finally succeed)...

Many of the gnu tools started life as BSD code that was hacked on and
rebranded with the GPL. Most of that original code is now gone, but in
the early days it was the source of much friction between the BSD and
GPL communities, even if a lot (all) of the code was eventually
replaced... It wasn't so much the use of the code that bothered
people, but the filing off of the original attributions...  All that's
water under the bridge, but the fact that this happened, as well as
many other incidents in the early days, goes a long way to explain
many of the hard feelings and out-sized reactions you used to see back
in the day.... This is also an important motivating factor for the
foundation that Linux was built on: This friction, the causes of which
were partially real or and partially imagined, drive much innovation
in both camps...

Warner

On Tue, Mar 14, 2017 at 10:20 AM, Arthur Krewat <krewat at kilonet.net> wrote:
> Everything. I'm trying to grok how long Linux as a whole was in active
> development. That includes all the GNU utilities, GCC, everything.
>
> Just like a "regular" corporate development environment would have devoted
> to the cause :)
>
>
>
>
> On 3/14/2017 11:57 AM, Michael Kjörling wrote:
>>
>> On 14 Mar 2017 11:51 -0400, from krewat at kilonet.net (Arthur Krewat):
>>>
>>> in reality, how much of Linux was based on previous works?
>>
>> Linux the kernel, or Linux the usable operating system (which would
>> include at least the essential userspace parts)?
>>
>

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Tony Finch]

Warner Losh <imp at bsdimp.com> wrote:

> Many of the gnu tools started life as BSD code that was hacked on and
> rebranded with the GPL. [context brutally snipped]

This brings up questions about how GNU and BSD operated around 1990ish.
I'm aware of Bostic's campaign to replace the AT&T code in BSD, which led
to the almost-completely-free Net/2. What I wonder is how much of this was
duplicating work also done under the GNU umbrella? How much of it was
authors donating their rewritten utilities to both projects? What was the
state of the GNU project when Bostic started his campaign?

Tony.
-- 
f.anthony.n.finch  <dot at dotat.at>  http://dotat.at/  -  I xn--zr8h punycode
Dover, Wight, Portland, Plymouth: Southwest 5 to 7, occasionally gale 8 except
in Plymouth. Moderate or rough. Fair then occasional rain. Good, occasionally
poor.

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Jeremy C. Reed]

> This brings up questions about how GNU and BSD operated around 1990ish.
> I'm aware of Bostic's campaign to replace the AT&T code in BSD, which led
> to the almost-completely-free Net/2. What I wonder is how much of this was
> duplicating work also done under the GNU umbrella? How much of it was
> authors donating their rewritten utilities to both projects? What was the
> state of the GNU project when Bostic started his campaign?

Have a look at the following:

GNU's Bulletin, vol. 1 no. 6, January, 1989
Contents of Beta Test Tape
https://www.gnu.org/bulletins/bull6.html#SEC17

GNU's Bulletin, vol. 1 no. 7, June, 1989
https://www.gnu.org/bulletins/bull7.html
"A collection of utilities for file manipulation, including ls, mv, cp, 
cat, rm, du, head, tail and cmp will be released soon."
...
"The GNU project is working to provide reimplementations of System V 
features that Berkeley Unix lacks, such as improved shells and make 
commands."

GNU's Bulletin, vol. 1 no. 9, June 1990
https://www.gnu.org/bulletins/bull9.html#SEC10
GNU Project Status Report
"We have added a collection of utilities for file manipulation to the 
Pre-Release tape. The collection includes ls, mv, cp, cat, rm, du, head, 
tail, cmp, chmod, mkdir, and ln."

So around same time GNU project didn't publish some the most common 
tools, but soon did. I didn't check, but I am pretty sure these are all 
different code than the rewritten BSD code. Duplicated work.

[TUHS] GNU vs BSD before the lawsuit and before Linux

[arnold]

"Jeremy C. Reed" <reed at reedmedia.net> wrote:

> So around same time GNU project didn't publish some the most common 
> tools, but soon did. I didn't check, but I am pretty sure these are all 
> different code than the rewritten BSD code. Duplicated work.

ISTR that the smaller utils were duplicated. 4.4BSD shipped gawk instead
of original Unix awk, and used GCC (and I guess the binutils) as the
compiler suite.  So some GNU stuff was used.

Arnold

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Wesley Parish]

If you read the early GNUs Bulletins you find a quite positive attitude towards the BSD community.

Wesley Parish

Quoting arnold at skeeve.com:

> "Jeremy C. Reed" <reed at reedmedia.net> wrote:
> 
> > So around same time GNU project didn't publish some the most common 
> > tools, but soon did. I didn't check, but I am pretty sure these are
> all 
> > different code than the rewritten BSD code. Duplicated work.
> 
> ISTR that the smaller utils were duplicated. 4.4BSD shipped gawk
> instead
> of original Unix awk, and used GCC (and I guess the binutils) as the
> compiler suite. So some GNU stuff was used.
> 
> Arnold
>  



"I have supposed that he who buys a Method means to learn it." - Ferdinand Sor,
Method for Guitar

"A verbal contract isn't worth the paper it's written on." -- Samuel Goldwyn

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Warner Losh]

On Sun, Mar 19, 2017 at 1:18 AM,  <arnold at skeeve.com> wrote:
> "Jeremy C. Reed" <reed at reedmedia.net> wrote:
>
>> So around same time GNU project didn't publish some the most common
>> tools, but soon did. I didn't check, but I am pretty sure these are all
>> different code than the rewritten BSD code. Duplicated work.
>
> ISTR that the smaller utils were duplicated. 4.4BSD shipped gawk instead
> of original Unix awk, and used GCC (and I guess the binutils) as the
> compiler suite.  So some GNU stuff was used.

All the GNU and X11 stuff was under contrib in 4.4-lite. This included
gawk, gcc, binutils, perl, emacs, flex, gdb, groff, kermit, libg++,
mh, nvi, rcs, gnu sort and a few other sundries. But the research awk
was also included. The build system by default included gawk though...

Warner

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Ron Natalie]

I could never convince the principals to call the FreeBSD project "Radio
Free Berkeley."   As for duplication of effort, I'm not sure anybody cared.
Certainly RMS didn't give a rats ass.
I suspect some of the stuff came from sources outside of both projects, like
stuff we did at BRL (although, those using that tape need to be careful,
most of that stuff came right out of the system V sources, hacked over to
work on BSD).

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1796 bytes --]

Below...  I agree!!!

On Tue, Mar 14, 2017 at 11:38 AM, Larry McVoy <lm at mcvoy.com> wrote:

>
> As someone who dedicated a bunch of his life to Unix, it pains me to say
> it but Linux is better than a lot of the Unix systems from back in the
> day.  I loved SunOS but I wouldn't trade today's Linux for SunOS and I
> don't think there are very many people who would disagree.
>
> It got better than "good enough".


​It doesn't pain me at all.   It is exactly what Prof. Christiansen says
will happen... the "good enough" (but disruptive) technology is on faster
growth than the "better" (but sustaining) one.   At some point that curves
will cross and what was once "good enough" now starts to control the
market.   That is what happened here.  Linux  is the "better" technology
for some value of "better." ​

The term "disruptive technology" gets tossed around a lot, but very few
people have actually read his book.  It really is an wonderful read.  He
nails our business.   The point is when the tech is birthed >>new group of
people<< don't care that the new technology is not as good - they value it
for some other reason.  One reason is that it is always less expensive.
Plus because it is a "lessor" technology, the "market leaders" have been
taught by the Harvard Business School and to ignore the lessor as a distraction
(no margins, they can not compete, they don't do the same things as us...
you have heard it all).   But that group of "few people" grows and the
value of the lessor tech to them, out weights, the limitations and they
provide the cash/incentive to make it better and it does get better.

Clem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170314/10158197/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[tfb]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2075 bytes --]

On 14 Mar 2017, at 14:43, Clem Cole <clemc at ccc.com> wrote:
> 
> I'm curious for the Banks, in your experience - which were the UNIX vendors that were pushing 9-figure UNIX boxes.  I'll guess, IBM was one of them.  Maybe NCR.     What HP, Sun, DEC in that bundle?

I didn't mean 9-figure sums on single machines: I meant that much for an estate.  Typically companies would have machines from more than one vendor: where I was we had IBM, HP, Sun in the Unix estate at least.  Then based on a fully-stuffed high-end machine costing ~$1M (which is about right), you need 100 to be 9 figures.  Where I was we had 25 top-end machines from the vendor I knew best I think, and probably as many again from each the two others, as well as a bunch (low thousands I think) of lesser machines.

> I ​believe that the point you are making is that "white box" PC's running a UNIX-like system - aka Linux could comes pretty close to doing what the highly touted AIX, NCR et al were doing and were "good enough" to get the job done.

No, not really: what I'm saying is that the deployments of big expensive Unix systems were *not* blazingly successful (for reasons which may or may not have had to do with Unix, and which I believe mostly but not entirely did not in fact), and the people who sign off that kind of purchase then have the 'Unix bad' bit set, and so anything which is being pushed as *not* Unix smells like good to them.

There is  no particular reason to think that what they are doing now will work any better, other than that I think it's obvious by now that the huge-gold-plated-machine idea doesn't work very well (with the possible exception of z series, which is not Unix of course), and much smaller silver-plated systems are just better and also offer stupidly more bang per buck.  Also they have probably learned some lessons from the first iteration so less dumb mistakes will be made.

--tim
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170314/fdfe9141/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Josh Good]

On 2017 Mar 14, 10:43, Clem Cole wrote:
> On Tue, Mar 14, 2017 at 7:35 AM, Tim Bradshaw <tfb at tfeb.org> wrote:
> 
> >  Linux is not Unix, and runs on cheap tin.
> >
> I believe that
> the point you are making is that "white box" PC's running a UNIX-like
> system - aka Linux could comes pretty close to doing what the highly touted
> AIX, NCR et al were doing and were "good enough" to get the job done.

Well, an HP Proliant (or Dell or Lenovo, etc.) machine, with its 
hardware-RAID battery-backed hard disk controller, redundant power supplies, 
lights-out remote access to firmware/BIOS, and 512 GB or more of RAM, is not 
exactly a "white box" PC - although it is undoubtely PC-based. Those things 
are mass-produced for the Windows market, but run Linux just the same.

If that system can be had, with Linux and full or source code, for 20% of 
the cost of a similar "highly touted" AIX or HP/UX or SPARC machine... 
well, that's pretty much a game over situation for several formerly 
incumbent UNIX-branded vendors.

> And that's not a statement about UNIX as much as a statement about, the
> WINTEL ecosystem, that Linux sat on top of and did an extremely impressive
> job of utilizing.

Totally agree. But it's also a statement about how when UNIX (the by hackers, 
for the hackers, operating system) closed its source code, it signed its 
future, unappealable, certain demise.

In "internet lingo": UNIX closed its source, that was felt as breakage, and
it was "routed around". Therefore, Linux.

Fellow list member Larry McVoy shaw it comming from the very beginning, he
has a paper about it: http://www.landley.net/history/mirror/unix/srcos.html

-- 
Josh Good

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 3411 bytes --]

On Tue, Mar 14, 2017 at 6:45 PM, Josh Good <pepe at naleco.com> wrote:

> In "internet lingo": UNIX closed its source, that was felt as breakage, and
> it was "routed around". Therefore, Linux.
>

​Arghhh ​..  sorry you pushed a "Clem Hot Button" -- Traditional AT&T UNIX
was always "Open Source" but it was licensed for $s however and you could
see it unless you paid to get a ticket from AT&T.  But if you did, it was
very open and if fact that is why Unix flourished.  I admit, I have found
it fascinating to hear from many of you that you did not believe it was so
easy to get access to the sources in those days, but the truth is - it was
open and because it was open, an industry was born around UNIX.   Not the
real "closed" systems of the day.

That said... the original code was never "Free and Open Source" - although
some believe it was made so (such as the UCB lawyers described here and
I'll not re-debate).   Others on this list, such as Larry, strongly believe
that IP was ripped off and have argued that.  The argument about
BSDi/386BSD/ et al is based is the "Free" part, not the "Open" part.

As I have pointed out, I switched from 386BSD to Linux because I was
worried we were going to lose access to "free" UNIX.  This is very much the
same as what happened by many other hackers in the day.  And as Larry
points out, when people like Larry, me et al started hacking, Linux
improved.   But the improvement happened *because of the economics of the
system*.

The other issue is that economics of UNIX changed.  When UNIX was
originally developed, for a University, the cost the systems was say
50-100K and the cost of Unix was at most $100.  For a firm it was $20K for
the first system and 5K for each system there after.   Expensive, but
manageable.

As for the cost of entry in a Unix system dropped to a $5-$10K for the HW
and SW together, the AT&T went up to $100K for the first system and $20K
for the second, although if got a redistribution license it could drop to
about $1K per system (BTW - that was the big fight with MSFT during the
time of the negotiation for what would become the System III license -
Gates wanted to pay $25 per CPU for Xenix and we laughed at by AT&T, DEC,
HP, IBM et al - I was in the room during that discussion in fact).

My point is that you (and many others)  equate "open" and "free" - I ask
you to please not make that error.   Open means we can talk about it and
share it, see it.  Which is exactly what we did "back in the day".  But as
people pointed out you had to pay AT&T to be a member of the UNIX club if
you were commercial, although any University type could be apart for free.
  Note the university restrictions were imposed by them (for probably good
reasons), but none the less - the code was open and >>was<< made available
at many places.

But it was not "free."  Linus chose to make make Linux free, bless him.
This single act, changed the economic potential of his "product" and in the
end, is what allowed it to expand.

But as I said, earlier today, this is right out Prof. Clay Christensen's
disruptive technology theory
<https://www.amazon.com/Innovators-Dilemma-Technologies-Management-Innovation/dp/1633691780/ref=dp_ob_title_bk>
.

Rant over....

Clem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170314/0b28e5bb/attachment-0001.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[arnold]

Clem Cole <clemc at ccc.com> wrote:

> But the improvement happened *because of the economics of the system*.

I guess that's true. My experience is that GNU Awk (gawk) was used on
the side a lot, alongside Unix awk on Unix systems.  When gawk became
*the* awk on Linux systems, that's when it really started getting pounded
on, and that's when many of the significant bugs and/or performance
issues got shaken out.

I suspect that this is true of the other major GNU tools, such as
Bash, coreutils, etc.

Arnold

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Josh Good]

On 2017 Mar 14, 21:11, Clem Cole wrote:
> 
> My point is that you (and many others)  equate "open" and "free" - I ask
> you to please not make that error.   Open means we can talk about it and
> share it, see it.  Which is exactly what we did "back in the day".  But as
> people pointed out you had to pay AT&T to be a member of the UNIX club if
> you were commercial, although any University type could be apart for free.

What UNIX for PC in the '90s had the option to buy a source code license
for that specific version, so that PC hackers could write drivers for
their hardware and tune the kernel internals to their liking, or be able
to fix themselves a bug in the serial port driver, etc.?

Certainly not OpenServer, not UnixWare nor SCO Xenix. Did DELL Unix
offered a payware source code license for their product? I'm not aware
of such.

From System V onwards, UNIX became closed source in what matters, that
is, the version running on your hardware and the version with the drivers
you are using (unless you were an employee at IBM, DEC, HP or SUN running
propietary hardware and happened to be in the right group).

It is obvious to me that RMS's GNU movement was aimed at solving that
very problem. And if that was a problem, then the "UNIX openness" you
talk about does not seem to have been very practical at all. At least,
it was totally useless to PC hackers, like Linus Torvalds - he had to
write his own UNIX, because he was not able to get any UNIX source code
he could readily compile and run on his i386.

-- 
Josh Good

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

SVR4

On Wed, Mar 15, 2017 at 3:28 PM, Josh Good <pepe at naleco.com> wrote:

> On 2017 Mar 14, 21:11, Clem Cole wrote:
> >
> > My point is that you (and many others)  equate "open" and "free" - I ask
> > you to please not make that error.   Open means we can talk about it and
> > share it, see it.  Which is exactly what we did "back in the day".  But
> as
> > people pointed out you had to pay AT&T to be a member of the UNIX club if
> > you were commercial, although any University type could be apart for
> free.
>
> What UNIX for PC in the '90s had the option to buy a source code license
> for that specific version, so that PC hackers could write drivers for
> their hardware and tune the kernel internals to their liking, or be able
> to fix themselves a bug in the serial port driver, etc.?
>
> Certainly not OpenServer, not UnixWare nor SCO Xenix. Did DELL Unix
> offered a payware source code license for their product? I'm not aware
> of such.
>
> From System V onwards, UNIX became closed source in what matters, that
> is, the version running on your hardware and the version with the drivers
> you are using (unless you were an employee at IBM, DEC, HP or SUN running
> propietary hardware and happened to be in the right group).
>
> It is obvious to me that RMS's GNU movement was aimed at solving that
> very problem. And if that was a problem, then the "UNIX openness" you
> talk about does not seem to have been very practical at all. At least,
> it was totally useless to PC hackers, like Linus Torvalds - he had to
> write his own UNIX, because he was not able to get any UNIX source code
> he could readily compile and run on his i386.
>
> --
> Josh Good
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/62f36405/attachment-0001.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Ron Natalie]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 613 bytes --]

We had the source code from Interactive Systems for their 386 implementations of IS/1 (pretty bare bones SysV)

Back around 1988.   I had it running on the PC and also was porting it to an Multibus II system (message passing coprocessor).

 

Anyhow you didn’t need the source to rebuild the kernels and write device drivers.   The necessary header file were there along

with the .o files needed to link your new stuff to the kernel.

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/776a07b2/attachment-0001.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

['Josh Good']

On 2017 Mar 15, 16:26, Ron Natalie wrote:

> We had the source code from Interactive Systems for their 386
> implementations of IS/1 (pretty bare bones SysV)

You had it. But was it available as a purchasing option "through the
distribution channel" to the public at large, or was it only available
to companies which entered a "joint venture" with Interactive Systems?

> Anyhow you didn???t need the source to rebuild the kernels and write
> device drivers.  The necessary header file were there along with the
> .o files needed to link your new stuff to the kernel.

If you wanted to fix the serial driver to work with a faster UART chip,
you could not do it (other than hacking hex in the object files).
Rebuilding the kernel pretty much only existed because the kernel used
hard coded config settings which needed relinking to be changed. The
object files themselves were set in stone by the vendor and the final
user had to option to change them.

If relinking the kernel to load different kernel modules, and the
ability to write device drivers for certain subsystems equals an open
system, then Windows NT is also an open system.

-- 
Josh Good

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

Sorry -- finger fumble...

SVR4 (aka UnixWare) was available for source  - the problem is many people
did like the price to see it.   It was $100K.  But the source was available
it was open and many, many of people with PC and had access to it, wrote
drivers for it etc.  There were books published about it.  It was hardly
secret.

That said, BSDi was $1K until the law suite and the pretty reasonable at
the time, and NET2 would eventually become free in the same way as Linux -
purely a copying fee.

Linus has gone on record if he had know about the 386BSD download, he would
have used it. It was a case of not knowing.   But as Larry points out, some
people still are not happy with the results.  It's also not clear that
people like me would still not gotten scared when the court case came --
which clouded things...

I'm not sating Linux was (and is not) important.

Just saying please don't say UNIX was not Open.   It was.  Unix was not
Free.

On Wed, Mar 15, 2017 at 3:28 PM, Josh Good <pepe at naleco.com> wrote:

> On 2017 Mar 14, 21:11, Clem Cole wrote:
> >
> > My point is that you (and many others)  equate "open" and "free" - I ask
> > you to please not make that error.   Open means we can talk about it and
> > share it, see it.  Which is exactly what we did "back in the day".  But
> as
> > people pointed out you had to pay AT&T to be a member of the UNIX club if
> > you were commercial, although any University type could be apart for
> free.
>
> What UNIX for PC in the '90s had the option to buy a source code license
> for that specific version, so that PC hackers could write drivers for
> their hardware and tune the kernel internals to their liking, or be able
> to fix themselves a bug in the serial port driver, etc.?
>
> Certainly not OpenServer, not UnixWare nor SCO Xenix. Did DELL Unix
> offered a payware source code license for their product? I'm not aware
> of such.
>
> From System V onwards, UNIX became closed source in what matters, that
> is, the version running on your hardware and the version with the drivers
> you are using (unless you were an employee at IBM, DEC, HP or SUN running
> propietary hardware and happened to be in the right group).
>
> It is obvious to me that RMS's GNU movement was aimed at solving that
> very problem. And if that was a problem, then the "UNIX openness" you
> talk about does not seem to have been very practical at all. At least,
> it was totally useless to PC hackers, like Linus Torvalds - he had to
> write his own UNIX, because he was not able to get any UNIX source code
> he could readily compile and run on his i386.
>
> --
> Josh Good
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/fb454390/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Larry McVoy]

On Wed, Mar 15, 2017 at 03:45:24PM -0400, Clem Cole wrote:
> SVR4 (aka UnixWare) was available for source  - the problem is many people
> did like the price to see it.   It was $100K.  But the source was available
> it was open and many, many of people with PC and had access to it, wrote
> drivers for it etc.  

That's a pretty peculiar definition of open.  Which is fine, I guess,
but you need to realize that that's open much like a high end country
club is open.  It's open to the rich people, to the connected people,
everyone else is left out in the cold.

In terms of source access, you're in the country club.  You are looking
around and you see all these other people in the club and that turns into
"many, many people" but it's not.  Millions of people, with the ability
to do something with the source, did not have access to the source.
$100K to someone with an ivy league education and a career that matched
may have seemed fine.  What about some talented hacker in, say, Finland?

What the so-called open people didn't get is that there was all this
talent that could be harnessed, in many cases for free, if you gave
them source.  It's too easy to look at your walled garden and see all
your friends there and go "everything was fine".  It wasn't, and as Josh
said, the world "routed around" the problem.  Which sort of proves it
was a problem.

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

My point is that is was on the *price list, *the sources were never hidden
away.   And that a lot of people did have access to it.  Your point - the
prices to get a ticket was too high and thus, when the price was even less,
even more had access.    Which I did not (and do not) disagree.

But Unix was open, people did discussed it, people did look at it, learned
from it etc....  that was not true of "closed systems" like say Cisco's.
Our even VMS, although VM, TSS, OS/360 and the like were "Open."   That's
why we have a UNIX "industry" -- it spread beyond the "ivy league" as you
said it.   The ideas leaked out, because AT&T made it open - because they
had by the 1956 consent decree et al....

That is a clear distinction.  And please its not about a wall garden ...
because it really was not that hard.  I'm not disagreeing that it did not
happen and your point is that people were excluded ... I get that.   But
don't call Unix closed because there was a price (aka a ticket).   It just
was not "free" -- that's all I'm saying and as you have pointed out that
difference was in practice to many, many people large (which I'm not
disagreeing).


You and I really need to have the beer together ;-)

On Wed, Mar 15, 2017 at 4:27 PM, Larry McVoy <lm at mcvoy.com> wrote:

> On Wed, Mar 15, 2017 at 03:45:24PM -0400, Clem Cole wrote:
> > SVR4 (aka UnixWare) was available for source  - the problem is many
> people
> > did like the price to see it.   It was $100K.  But the source was
> available
> > it was open and many, many of people with PC and had access to it, wrote
> > drivers for it etc.
>
> That's a pretty peculiar definition of open.  Which is fine, I guess,
> but you need to realize that that's open much like a high end country
> club is open.  It's open to the rich people, to the connected people,
> everyone else is left out in the cold.
>
> In terms of source access, you're in the country club.  You are looking
> around and you see all these other people in the club and that turns into
> "many, many people" but it's not.  Millions of people, with the ability
> to do something with the source, did not have access to the source.
> $100K to someone with an ivy league education and a career that matched
> may have seemed fine.  What about some talented hacker in, say, Finland?
>
> What the so-called open people didn't get is that there was all this
> talent that could be harnessed, in many cases for free, if you gave
> them source.  It's too easy to look at your walled garden and see all
> your friends there and go "everything was fine".  It wasn't, and as Josh
> said, the world "routed around" the problem.  Which sort of proves it
> was a problem.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/594d3ca0/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Arthur Krewat]

Another common thread in this discussion has been universities had 
access to the source code, so just go ask the right people at your school.

Which is all well and good, except for those that never went - for 
various reasons. I never even graduated High School ...

I had "access" to the 4.2/4.3BSD sources, but only because a friend of 
mine worked at a university.

It was when I was forewarned that said university threw away piles of 
VAXes and tapes that I then had my grubby hands on the source code to 
BSD and a few other things.

This is where the NFS 2.0 source code came from that I gave to Warren :)

When I got my grubby hands on FreeBSD, I nearly cried. This was before I 
went dumpster diving.


On 3/15/2017 4:27 PM, Larry McVoy wrote:
> SNIP
> In terms of source access, you're in the country club.
>
> SNIP

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

Arthur's behavior, I think really proves my point....   you were able to do
that because UNIX was (is) open and was being studies and used by a lot of
people had folks did have access to the source. You did not see VMS, NT or
the like having academic books written about them, and used as models in OS
classes.  You would not have been able to do that with NT or VMS.

And whole point is that when AT&T did try to pull it back as a trade
secret, they failed.  The courts said -- no.  This is open....  this has
been published - this story has been told.  These ideas out "out there."

That is why if AT&T had >>won<< the BSDi/UCB case, technical Linux and all
the clone were in violation also.   It it was found to be a "secret" - but
the court made it clear.  It was not.  And by the way, the Lions book could
never be legally bought - it was the most copied text around.

The whole term "Open Systems" was used to define Unix to being with.   The
idea is that Unix was the first of OS that people really knew what was
happening under the covers.  There were no secrets - you got / could get
the sources (even with Sys V) but it might cost $s.


I also understand the disappointment that many of you had because you did
not have access to the sources.  I get it.  I see that would have been
frustrating.   Particularly when it was so close and yet so far.  And it
does seem like it was club that you couldn't belong, which I find sad
because many of never looked at that way.   I also see that if you were not
in that club, you might be seen by some "in the club" as not having done
anything "worthwhile" and need to "prove" something.

If that is what how you feel, I truly want to make sure you understand I do
not believe either of those things and never have.



Josh -- all I am asking is you to be respectful of the term and the folks
that created it, industry and frankly the market and opportunity that Linux
and today's tech has so wonderful exploited.  So I ask you to please call
it Free and Open and I'm fine either way, although it will grate me when I
see you and other make that misunderstanding.

I believe that Linux was and >>is<< important and it does matter!!!  The
Cool Kids did something, I am many others are thankful and proud of them
for doing ... it Linux is an great piece of technology and its paying a lot
of bills for a lot of people today (including me).  I do not want to be
seen and knocking Linux in anyway.

But I do see a lot of people "knocking" Unix because it was not "free" and
frankly it was a different world.   That's why I'm trying help explain the
difference.

Maybe it is too subtle for you to see and you had to live it to fully
understand it.

I fear you think because I did have access to sources, I think I was some
how special.  My point has always been, we really were not.  Other than the
specialness of the time was based on economics, because the cost of the
systems that ran UNIX was so high, that was what limited.    Which comes
back to my main thesis...  this is an argument about economics and cost --
WINTEL economics changed things -- so the question is asked -- did having
access to the sources play into the openness or not?

My point is that it was open >>before<< WINTEL existed, so you can change
it being open or not.   What "Free UNIX" did was make the "wisdom" spread
even faster - it was an accelerant but it did not change the basic piece.

UNIX was different .. it was open... it open up minds and created and
industry, which now Linux (a "free" UNIX implementation) is undisputed
leader - created by some cool folks that  I personally very much respect
and admire.

I do fear a problem is that you seem to be equating "open" with "having
access to the source" - where as the term was coined to mean "the ideas are
available for all to see and share in" - as in a mathematical, and academic
style of openness. Open University, Open Book, Open Ideas etc...   I
suspect your definition has narrowed that definition to include that unless
the sources are in front of you, *the ideas are not really available; *which
is why I cringe and it is a hot button for me and I find it wrong.

In the end, definition does not change the status of what Unix was.   It
was the definition of Open Systems -- it was published and I do stand
behind that.   And in the end, it could not be claimed as trade secret
because it was ->> by definition<<- open and known. But traditional Unix
from AT&T was never >>free<< and that fact is not going to change either.
It may some how in the future, but that past is true and as a result, Linus
and other did an end-around and created and awesome >>free<< solution.


Clem





On Wed, Mar 15, 2017 at 7:46 PM, Arthur Krewat <krewat at kilonet.net> wrote:

> Another common thread in this discussion has been universities had access
> to the source code, so just go ask the right people at your school.
>
> Which is all well and good, except for those that never went - for various
> reasons. I never even graduated High School ...
>
> I had "access" to the 4.2/4.3BSD sources, but only because a friend of
> mine worked at a university.
>
> It was when I was forewarned that said university threw away piles of
> VAXes and tapes that I then had my grubby hands on the source code to BSD
> and a few other things.
>
> This is where the NFS 2.0 source code came from that I gave to Warren :)
>
> When I got my grubby hands on FreeBSD, I nearly cried. This was before I
> went dumpster diving.
>
>
> On 3/15/2017 4:27 PM, Larry McVoy wrote:
>
>> SNIP
>> In terms of source access, you're in the country club.
>>
>> SNIP
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/6b2c4939/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Steve Nickolas]

"Free" and "open" are, by themselves, ambiguous words.  It is when you 
combine them that you get the meaning that is often intended by Linux, GNU 
and OpenBSD people.

-uso.

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Ron Natalie]

I tripped across a project the other day which was a commercial,
proprietary, and undocumented system that still used "Open" in their company
name.
I'm trying to figure out what aspect they are claiming is open.

-----Original Message-----
From: TUHS [mailto:tuhs-bounces@minnie.tuhs.org] On Behalf Of Steve Nickolas
Sent: Wednesday, March 15, 2017 9:27 PM
To: Clem Cole
Cc: TUHS main list
Subject: Re: [TUHS] System Economics (was is Linux "officially branded
UNIX")

"Free" and "open" are, by themselves, ambiguous words.  It is when you
combine them that you get the meaning that is often intended by Linux, GNU
and OpenBSD people.

-uso.

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Charles Anthony]

On Wed, Mar 15, 2017 at 8:09 PM, Ron Natalie <ron at ronnatalie.com> wrote:

> I tripped across a project the other day which was a commercial,
> proprietary, and undocumented system that still used "Open" in their
> company
> name.
> I'm trying to figure out what aspect they are claiming is open.
>
>
I once saw a fiber optic connection box labeled something like "Open
Systems Interconnect" with a key access lock on it.

-- Charles
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/32a4762a/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Dan Cross]

On Wed, Mar 15, 2017 at 8:45 PM, Clem Cole <clemc at ccc.com> wrote:

> [...]
> In the end, definition does not change the status of what Unix was.   It
> was the definition of Open Systems -- it was published and I do stand
> behind that.   And in the end, it could not be claimed as trade secret
> because it was ->> by definition<<- open and known. But traditional Unix
> from AT&T was never >>free<< and that fact is not going to change either.
> It may some how in the future, but that past is true and as a result, Linus
> and other did an end-around and created and awesome >>free<< solution.
>
[...]


Hmm, this is quite interesting, but I had different impression of the
definition of "open" at the time: it seemed like what people were saying
when they said that Unix was "open" was much less about the source code,
but rather about the interfaces and APIs; in particular especially after
the standards bodies got together and starting writing down how things were
supposed to work. This led to vendor independence (to some extent, anyway)
and was a distinction from closed systems which were defined by a single
vendor who controlled everything about them (though presumably modulated by
customer demand), including the OS (since this was usually written in-house
for each platform. This even makes historical sense: Unix was written by a
third party who didn't design the hardware).

Consider DEC: In 1981, they had at least three hardware platforms intended
for the timesharing market, each running multiple operating systems: PDP-11
running RSX-11*, RT-11, RSTS/E and Ultrix-11 (Unix); PDP-10 running TOPS-10
and TOPS-20; VAX running VMS and Ultrix-32 (Unix). And this isn't to
mention any of the other stuff they were selling/supporting (PDP-8's, etc).
Of those software systems it's easy to see what Ultrix-11 and Ultrix-32
have in common; one has a reasonable shot at getting software written for
one running on the other. Contrast with RT-11 and VMS, or even RT-11 and
RSX. Similarly with IBM, CDC, HP, GE, etc.

In other words, the "openness" in "open systems" wasn't about code *for the
system itself*; it was about freedom from software lock-in to a particular
hardware vendor. Or, perhaps, openness to multiple system vendors
supporting the same customer-written code. "Open" in a sense closer to what
we now call "open source" (meaning the source was available for inspection)
came much later.

        - Dan C.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/e92167db/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[arnold]

Dan Cross <crossd at gmail.com> wrote:

> Hmm, this is quite interesting, but I had different impression of the
> definition of "open" at the time: it seemed like what people were saying
> when they said that Unix was "open" was much less about the source code,
> but rather about the interfaces and APIs;

Yes!!!! Portability of application code was a big issue, and the option
to avoid vendor lock-iin.

> In other words, the "openness" in "open systems" wasn't about code *for the
> system itself*; it was about freedom from software lock-in to a particular
> hardware vendor. Or, perhaps, openness to multiple system vendors
> supporting the same customer-written code.

You've hit the nail very much on the head.

This did come as a result of the "openness" that Clem is describing:
since people knew how "UNIX" (as a concept) worked, it was possible to
transfer both your source code, and your peoples' how-to-use-it knowledge
from one vendor to another.  This caused vendors to start competing
more on price / performance.

Arnold

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Arthur Krewat]

On 3/15/2017 8:45 PM, Clem Cole wrote:
> Arthur's behavior, I think really proves my point....   you were able 
> to do that because UNIX was (is) open and was being studies and used 
> by a lot of people had folks did have access to the source. You did 
> not see VMS, NT or the like having academic books written about them, 
> and used as models in OS classes. You would not have been able to do 
> that with NT or VMS.
>

Believe it or not, somewhere I believe I have microfiche of VMS 4.0 
source code somewhere. Which reminds me, I need to do something with that.

It came from the same place as the other source code I have.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170316/fd4ed1bd/attachment-0001.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[William Pechter]

VMS source fiche was very common of sites owned by large corporations.  Their IT  staff used it to research bugs... and as sample code for writing their own drivers etc... 


-----Original Message-----
From: Arthur Krewat <krewat@kilonet.net>
To: Clem Cole <clemc at ccc.com>
Cc: TUHS main list <tuhs at minnie.tuhs.org>
Sent: Thu, 16 Mar 2017 8:53
Subject: Re: [TUHS] System Economics (was is Linux "officially branded UNIX")



On 3/15/2017 8:45 PM, Clem Cole wrote:
> Arthur's behavior, I think really proves my point....   you were able 
> to do that because UNIX was (is) open and was being studies and used 
> by a lot of people had folks did have access to the source. You did 
> not see VMS, NT or the like having academic books written about them, 
> and used as models in OS classes. You would not have been able to do 
> that with NT or VMS.
>

Believe it or not, somewhere I believe I have microfiche of VMS 4.0 
source code somewhere. Which reminds me, I need to do something with that.

It came from the same place as the other source code I have.

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Josh Good]

On 2017 Mar 15, 20:45, Clem Cole wrote:

> Josh -- all I am asking is you to be respectful of the term and the folks
> that created it, industry and frankly the market and opportunity that Linux
> and today's tech has so wonderful exploited.

Clem, I am respectful of you and of all the list members from whose
experience and direct contact with "primordial Unix" I try to constantly
learn.

I, however, try to express opinion on the "openness" of Unix (post
V7). Nobody has been able to write a Unix from scratch without having had
access to the Unix source code: Tannenbaum had access to the Unix sources
before writing Minix, Linus had access to Minix source before writing
Linux, and in Dennis Ritchie's opinion Coherent was a "rewritten Unix"
done with the Unix sources printed next to the keyboard ("some parts
were written with our source nearby, but at least the effort had been
made to rewrite").

GNU rewrote all the surrounding Unix tools from scratch, that's true,
but they could not a kernel make.

So much for "openness of concepts".

It's the source that matters. Anything else, is ivory-towerism.

Post Lions' book being forbidden, Unix can boast little openness.

> I do fear a problem is that you seem to be equating "open" with "having
> access to the source" - where as the term was coined to mean "the ideas are
> available for all to see and share in" - as in a mathematical, and academic
> style of openness.

Regards,

-- 
Josh Good

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Chet Ramey]

On 3/15/17 4:27 PM, Larry McVoy wrote:
> On Wed, Mar 15, 2017 at 03:45:24PM -0400, Clem Cole wrote:
>> SVR4 (aka UnixWare) was available for source  - the problem is many people
>> did like the price to see it.   It was $100K.  But the source was available
>> it was open and many, many of people with PC and had access to it, wrote
>> drivers for it etc.  
> 
> That's a pretty peculiar definition of open.  Which is fine, I guess,
> but you need to realize that that's open much like a high end country
> club is open.  It's open to the rich people, to the connected people,
> everyone else is left out in the cold.

This is the same access vs. affordability argument we're seeing played out
in other segments of US society.

I have access to a Porsche, as do thousands of others (some of whom choose
not to buy one), but I can't afford one.

-- 
``The lyf so short, the craft so long to lerne.'' - Chaucer
		 ``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, UTech, CWRU    chet at case.edu    http://cnswww.cns.cwru.edu/~chet/

[TUHS] System Economics (was is Linux "officially branded UNIX")

[William Pechter]

Chet Ramey wrote:
> On 3/15/17 4:27 PM, Larry McVoy wrote:
>> On Wed, Mar 15, 2017 at 03:45:24PM -0400, Clem Cole wrote:
>>> SVR4 (aka UnixWare) was available for source  - the problem is many people
>>> did like the price to see it.   It was $100K.  But the source was available
>>> it was open and many, many of people with PC and had access to it, wrote
>>> drivers for it etc.  
>> That's a pretty peculiar definition of open.  Which is fine, I guess,
>> but you need to realize that that's open much like a high end country
>> club is open.  It's open to the rich people, to the connected people,
>> everyone else is left out in the cold.
> This is the same access vs. affordability argument we're seeing played out
> in other segments of US society.
>
> I have access to a Porsche, as do thousands of others (some of whom choose
> not to buy one), but I can't afford one.
>
I have copies of the old 8086/8088 PC Xenix.  I don't know if SCO or
Microsoft even sold
the sources.  My old boss ran very early Microsoft SCO on a PDP11. 
Don't know if he had
sources.

Even working as a sysadmin for an AT&T oem of System V wouldn't get me
access to the source
through normal corporate means.  Perhaps I could've bribed an engineer
with a Pizza to let me get
to his machine without a screen lock... but he'd have been risking his job.

Bill

-- 
Digital had it then.  Don't you wish you could buy it now!
pechter-at-gmail.com  http://xkcd.com/705/

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Josh Good]

On 2017 Mar 15, 15:45, Clem Cole wrote:
> 
> SVR4 (aka UnixWare) was available for source  - the problem is many people
> did like the price to see it.   It was $100K.  But the source was available
> it was open and many, many of people with PC and had access to it, wrote
> drivers for it etc.  There were books published about it.  It was hardly
> secret.

Nobody says UNIX source code was "secret". It just was not open after
UNIX began to be directly sold by AT&T post Bell-breakage.

If UNIX source code was "open" at $100K, then Windows NT source code can
also be seen as open if you have enough money to buy Microsoft.

> Just saying please don't say UNIX was not Open.   It was.  Unix was not
> Free.

I beg to differ. UNIX stopped being open when the Lion's book could not
be legally sold anymore at bookstores. That happened even earlier than
System V, it happened when AT&T released V7. The reason that AT&T stated
was that they wanted to keep "UNIX source code" as a "trade secret".

So this begs the question: how can something, anything, be at the same
time "open" and a "trade secret"?

No doubt some argumentation can be concocted to marry both concepts, but
I have that feeling it's going to be a hard one to swallow.

To me, open means libre access, because if there is no libre access, then
it is what is known as closed.

Please note that libre access --when applied to source code-- does not
necessarily mean "up for grabs and redistribution" - it just means libre
access to random eyeballs in meatspace.

-- 
Josh Good

[TUHS] System Economics (was is Linux "officially branded UNIX")

[William Pechter]

Josh Good wrote:
> On 2017 Mar 15, 15:45, Clem Cole wrote:
>> SVR4 (aka UnixWare) was available for source  - the problem is many people
>> did like the price to see it.   It was $100K.  But the source was available
>> it was open and many, many of people with PC and had access to it, wrote
>> drivers for it etc.  There were books published about it.  It was hardly
>> secret.
> Nobody says UNIX source code was "secret". It just was not open after
> UNIX began to be directly sold by AT&T post Bell-breakage.
>
> If UNIX source codeHa was "open" at $100K, then Windows NT source code can
> also be seen as open if you have enough money to buy Microsoft.
Having worked at a minicomputer company (Concurrent Computer
Corporation) back in the 87 days... I can say that
there was no way I could access the SysV sources without being an
approved developer or support engineer.
I was the IT Systems Administrator with the company managing their Xelos
SysVR2 systems in the MIS department
and I had no access.

When I found a serious bug they looked at it.  Reproduced it.  Reported
it to AT&T who checked the will not fix box on
their ticket and closed the bug.  The problem was that cron would malloc
memory until it couldn't get any more and core
dump.  This  stopped automatic backups and jobs from being scheduled
reliably and was critical to my operations.

The fix was I had to write a script to kill -0 cron with a sleep... and
if the job was no longer there  -- restart cron.

AT&T support basically said "Get a machine that implements demand paged
virtual memory and it won't happen."

Pretty sad.  I had worked for DEC and other places that would've fixed
the code for a customer.  Especially an OEM.

Concurrent had pretty much put Unix on the back burner until they bought
Masscomp.   Their only OS they were pushing
at the time was OS/32.  Xelos was a pretty decent SVR5.2 port -- the
next version even had ksh in it.

I wanted to get a 3280 and see how Xelos on it compared with something
equivalent on a VAX 8650.

Bill
Pechter at gmail.com

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

BTW Josh, I am trying to be respectful here.  I suspect you are tad younger
I am and your early introduction into UNIX was on the WINTEL platform, not
on the DEC systems like mine.  So, if I'm going to make a guess you were
not in a position when you were introduced to be able to get access to the
sources.

So in your experience the UNIX source were closed to you personally (and
many others).  I get that.   But it does not change the fact it, there were
available and there open and were not a secret.   Which was very different
from many of "closed" systems (says Cisco, or much of the other
infrastructure) of the day.  They always have been.  Even System V.

It was quite easy to get source if you were willing (and could pay).   I'm
not suggesting that it was easy for you could and I understand that
frustration.  I personally would not have been able to pay for the
licenses, but I was being employed by firms that could and valued my
abilities, so they did.  This was also true for many educational
institutions.

Sun and DEC actually were quite liberal with their source licenses, because
AT&T had been.  They had to be also - because their customer required it.

The point is there is a difference between "open" and "free."   There are a
lot of things that are open and we can look at but not touch or have the
wearwithall to modify.  But that does not change their openness - we can
still (and do) learn from them.

Linus and many of us learned because UNIX (the ideas) and the basic
implementations were open.  We talked about them, they were well specified.
  We wrote application that relied on those ideas, APIs etc.   And Linus,
Andy Tannenbaum and Plaguer before them reimplemented those ideas and
created clones.   *Unix was and is "open" and the implementations were and
are available.*   The problem for many is the price to look at the
implementations - that I grant. And for many, for some of those
implementations, can be high.  But it does not make them "closed."

The effect may seem that way to you, but it was not and is not the same.
All, I'm asking you to say, is that traditional UNIX implementations such
as System V were not "Free and Open," unlike Linux some of the other Unix
clones.  And that make all the difference.

They were and still are open.

On Wed, Mar 15, 2017 at 3:28 PM, Josh Good <pepe at naleco.com> wrote:

> On 2017 Mar 14, 21:11, Clem Cole wrote:
> >
> > My point is that you (and many others)  equate "open" and "free" - I ask
> > you to please not make that error.   Open means we can talk about it and
> > share it, see it.  Which is exactly what we did "back in the day".  But
> as
> > people pointed out you had to pay AT&T to be a member of the UNIX club if
> > you were commercial, although any University type could be apart for
> free.
>
> What UNIX for PC in the '90s had the option to buy a source code license
> for that specific version, so that PC hackers could write drivers for
> their hardware and tune the kernel internals to their liking, or be able
> to fix themselves a bug in the serial port driver, etc.?
>
> Certainly not OpenServer, not UnixWare nor SCO Xenix. Did DELL Unix
> offered a payware source code license for their product? I'm not aware
> of such.
>
> From System V onwards, UNIX became closed source in what matters, that
> is, the version running on your hardware and the version with the drivers
> you are using (unless you were an employee at IBM, DEC, HP or SUN running
> propietary hardware and happened to be in the right group).
>
> It is obvious to me that RMS's GNU movement was aimed at solving that
> very problem. And if that was a problem, then the "UNIX openness" you
> talk about does not seem to have been very practical at all. At least,
> it was totally useless to PC hackers, like Linus Torvalds - he had to
> write his own UNIX, because he was not able to get any UNIX source code
> he could readily compile and run on his i386.
>
> --
> Josh Good
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/32eff3db/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Wesley Parish]

One of the problems was the hardware system price ratio to software system price. When you could get 
a 486 PC for $5-10k and the SysV source license (for 4.3BSD!) was $100k, it seemed rather monstrously 
disproportionate. :)

This mismatch didn't exist in the Minicomputer world, where a VAX cost rather more than $5-10k and 
the price for a source license was thus not disproportionate.

FWVLIW

Wesley Parish

Quoting Clem Cole <clemc at ccc.com>:

> BTW Josh, I am trying to be respectful here. I suspect you are tad
> younger
> I am and your early introduction into UNIX was on the WINTEL platform,
> not
> on the DEC systems like mine. So, if I'm going to make a guess you were
> not in a position when you were introduced to be able to get access to
> the
> sources.
> 
> So in your experience the UNIX source were closed to you personally
> (and
> many others). I get that. But it does not change the fact it, there
> were
> available and there open and were not a secret. Which was very
> different
> from many of "closed" systems (says Cisco, or much of the other
> infrastructure) of the day. They always have been. Even System V.
> 
> It was quite easy to get source if you were willing (and could pay).
> I'm
> not suggesting that it was easy for you could and I understand that
> frustration. I personally would not have been able to pay for the
> licenses, but I was being employed by firms that could and valued my
> abilities, so they did. This was also true for many educational
> institutions.
> 
> Sun and DEC actually were quite liberal with their source licenses,
> because
> AT&T had been. They had to be also - because their customer required
> it.
> 
> The point is there is a difference between "open" and "free." There are
> a
> lot of things that are open and we can look at but not touch or have
> the
> wearwithall to modify. But that does not change their openness - we can
> still (and do) learn from them.
> 
> Linus and many of us learned because UNIX (the ideas) and the basic
> implementations were open. We talked about them, they were well
> specified.
>  We wrote application that relied on those ideas, APIs etc. And Linus,
> Andy Tannenbaum and Plaguer before them reimplemented those ideas and
> created clones. *Unix was and is "open" and the implementations were
> and
> are available.* The problem for many is the price to look at the
> implementations - that I grant. And for many, for some of those
> implementations, can be high. But it does not make them "closed."
> 
> The effect may seem that way to you, but it was not and is not the
> same.
> All, I'm asking you to say, is that traditional UNIX implementations
> such
> as System V were not "Free and Open," unlike Linux some of the other
> Unix
> clones. And that make all the difference.
> 
> They were and still are open.
> 
> On Wed, Mar 15, 2017 at 3:28 PM, Josh Good <pepe at naleco.com> wrote:
> 
> > On 2017 Mar 14, 21:11, Clem Cole wrote:
> > >
> > > My point is that you (and many others) equate "open" and "free" - I
> ask
> > > you to please not make that error. Open means we can talk about it
> and
> > > share it, see it. Which is exactly what we did "back in the day".
> But
> > as
> > > people pointed out you had to pay AT&T to be a member of the UNIX
> club if
> > > you were commercial, although any University type could be apart
> for
> > free.
> >
> > What UNIX for PC in the '90s had the option to buy a source code
> license
> > for that specific version, so that PC hackers could write drivers for
> > their hardware and tune the kernel internals to their liking, or be
> able
> > to fix themselves a bug in the serial port driver, etc.?
> >
> > Certainly not OpenServer, not UnixWare nor SCO Xenix. Did DELL Unix
> > offered a payware source code license for their product? I'm not
> aware
> > of such.
> >
> > From System V onwards, UNIX became closed source in what matters,
> that
> > is, the version running on your hardware and the version with the
> drivers
> > you are using (unless you were an employee at IBM, DEC, HP or SUN
> running
> > propietary hardware and happened to be in the right group).
> >
> > It is obvious to me that RMS's GNU movement was aimed at solving that
> > very problem. And if that was a problem, then the "UNIX openness" you
> > talk about does not seem to have been very practical at all. At
> least,
> > it was totally useless to PC hackers, like Linus Torvalds - he had to
> > write his own UNIX, because he was not able to get any UNIX source
> code
> > he could readily compile and run on his i386.
> >
> > --
> > Josh Good
> >
> >
>  



"I have supposed that he who buys a Method means to learn it." - Ferdinand Sor,
Method for Guitar

"A verbal contract isn't worth the paper it's written on." -- Samuel Goldwyn

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Clem Cole]

Right and as Larry points out that led to a club mentality, which can see
would (in practice) make people interpret something in a different way than
really was the case or the intended case.

On Wed, Mar 15, 2017 at 8:46 PM, Wesley Parish <wes.parish at paradise.net.nz>
wrote:

> One of the problems was the hardware system price ratio to software system
> price. When you could get
> a 486 PC for $5-10k and the SysV source license (for 4.3BSD!) was $100k,
> it seemed rather monstrously
> disproportionate. :)
>
> This mismatch didn't exist in the Minicomputer world, where a VAX cost
> rather more than $5-10k and
> the price for a source license was thus not disproportionate.
>
> FWVLIW
>
> Wesley Parish
>
> Quoting Clem Cole <clemc at ccc.com>:
>
> > BTW Josh, I am trying to be respectful here. I suspect you are tad
> > younger
> > I am and your early introduction into UNIX was on the WINTEL platform,
> > not
> > on the DEC systems like mine. So, if I'm going to make a guess you were
> > not in a position when you were introduced to be able to get access to
> > the
> > sources.
> >
> > So in your experience the UNIX source were closed to you personally
> > (and
> > many others). I get that. But it does not change the fact it, there
> > were
> > available and there open and were not a secret. Which was very
> > different
> > from many of "closed" systems (says Cisco, or much of the other
> > infrastructure) of the day. They always have been. Even System V.
> >
> > It was quite easy to get source if you were willing (and could pay).
> > I'm
> > not suggesting that it was easy for you could and I understand that
> > frustration. I personally would not have been able to pay for the
> > licenses, but I was being employed by firms that could and valued my
> > abilities, so they did. This was also true for many educational
> > institutions.
> >
> > Sun and DEC actually were quite liberal with their source licenses,
> > because
> > AT&T had been. They had to be also - because their customer required
> > it.
> >
> > The point is there is a difference between "open" and "free." There are
> > a
> > lot of things that are open and we can look at but not touch or have
> > the
> > wearwithall to modify. But that does not change their openness - we can
> > still (and do) learn from them.
> >
> > Linus and many of us learned because UNIX (the ideas) and the basic
> > implementations were open. We talked about them, they were well
> > specified.
> >  We wrote application that relied on those ideas, APIs etc. And Linus,
> > Andy Tannenbaum and Plaguer before them reimplemented those ideas and
> > created clones. *Unix was and is "open" and the implementations were
> > and
> > are available.* The problem for many is the price to look at the
> > implementations - that I grant. And for many, for some of those
> > implementations, can be high. But it does not make them "closed."
> >
> > The effect may seem that way to you, but it was not and is not the
> > same.
> > All, I'm asking you to say, is that traditional UNIX implementations
> > such
> > as System V were not "Free and Open," unlike Linux some of the other
> > Unix
> > clones. And that make all the difference.
> >
> > They were and still are open.
> >
> > On Wed, Mar 15, 2017 at 3:28 PM, Josh Good <pepe at naleco.com> wrote:
> >
> > > On 2017 Mar 14, 21:11, Clem Cole wrote:
> > > >
> > > > My point is that you (and many others) equate "open" and "free" - I
> > ask
> > > > you to please not make that error. Open means we can talk about it
> > and
> > > > share it, see it. Which is exactly what we did "back in the day".
> > But
> > > as
> > > > people pointed out you had to pay AT&T to be a member of the UNIX
> > club if
> > > > you were commercial, although any University type could be apart
> > for
> > > free.
> > >
> > > What UNIX for PC in the '90s had the option to buy a source code
> > license
> > > for that specific version, so that PC hackers could write drivers for
> > > their hardware and tune the kernel internals to their liking, or be
> > able
> > > to fix themselves a bug in the serial port driver, etc.?
> > >
> > > Certainly not OpenServer, not UnixWare nor SCO Xenix. Did DELL Unix
> > > offered a payware source code license for their product? I'm not
> > aware
> > > of such.
> > >
> > > From System V onwards, UNIX became closed source in what matters,
> > that
> > > is, the version running on your hardware and the version with the
> > drivers
> > > you are using (unless you were an employee at IBM, DEC, HP or SUN
> > running
> > > propietary hardware and happened to be in the right group).
> > >
> > > It is obvious to me that RMS's GNU movement was aimed at solving that
> > > very problem. And if that was a problem, then the "UNIX openness" you
> > > talk about does not seem to have been very practical at all. At
> > least,
> > > it was totally useless to PC hackers, like Linus Torvalds - he had to
> > > write his own UNIX, because he was not able to get any UNIX source
> > code
> > > he could readily compile and run on his i386.
> > >
> > > --
> > > Josh Good
> > >
> > >
> >
>
>
>
> "I have supposed that he who buys a Method means to learn it." - Ferdinand
> Sor,
> Method for Guitar
>
> "A verbal contract isn't worth the paper it's written on." -- Samuel
> Goldwyn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170315/af366922/attachment.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Dave Horsfall]

On Wed, 15 Mar 2017, Josh Good wrote:

> It is obvious to me that RMS's GNU movement was aimed at solving that 
> very problem. And if that was a problem, then the "UNIX openness" you 
> talk about does not seem to have been very practical at all. At least, 
> it was totally useless to PC hackers, like Linus Torvalds - he had to 
> write his own UNIX, because he was not able to get any UNIX source code 
> he could readily compile and run on his i386.

Perhaps I'm confused (not uncommon) but I have distinct memories of having 
a source licence for my BSD/OS system on a 386...

-- 
Dave Horsfall DTM (VK2KFU)  "Those who don't understand security will suffer."

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Jason Stevens]

Well $999 would get you source..

https://c1.staticflickr.com/1/32/93939063_729b710163_z.jpg?zz=1

With more and more magazines of the era being scanned and put online, I should try to find the 1800itsunix...

On March 17, 2017 3:47:55 AM GMT+08:00, Dave Horsfall <dave at horsfall.org> wrote:
>On Wed, 15 Mar 2017, Josh Good wrote:
>
>> It is obvious to me that RMS's GNU movement was aimed at solving that
>
>> very problem. And if that was a problem, then the "UNIX openness" you
>
>> talk about does not seem to have been very practical at all. At
>least, 
>> it was totally useless to PC hackers, like Linus Torvalds - he had to
>
>> write his own UNIX, because he was not able to get any UNIX source
>code 
>> he could readily compile and run on his i386.
>
>Perhaps I'm confused (not uncommon) but I have distinct memories of
>having 
>a source licence for my BSD/OS system on a 386...
>
>-- 
>Dave Horsfall DTM (VK2KFU)  "Those who don't understand security will
>suffer."

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170317/a39a0fca/attachment-0001.html>

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Warner Losh]

On Thu, Mar 16, 2017 at 8:16 PM, Jason Stevens
<jsteve at superglobalmegacorp.com> wrote:
> Well $999 would get you source..
>
> https://c1.staticflickr.com/1/32/93939063_729b710163_z.jpg?zz=1
>
> With more and more magazines of the era being scanned and put online, I
> should try to find the 1800itsunix...

In June 1993, you could already get FreeBSD, NetBSD and 386BSD as well
as BSD 4.4-lite. I think Minix was also available and several other
'also ran' Unix clones of the era whose names have slipped from my
memory....

Most of the folks in this thread are lamenting the era before the Net2
release when nothing was available without some kind of encumbrance.
And they do have a point. Where you went to school mattered a lot for
how much access to the sources you could get.

But I was at a school that had a liberal source access policy. You
asked Mike and he told you where to find the source. :) Mike was the
director of the computer center, and he also told you not to release
it and it would be an expellable offense if you shared it or copied
off the servers. But at the time, I didn't have enough disk space on
my PC to do that.... and I always had dialin access to the encumbered
4.2BSD sources as well as the SunOS 3.x sources. But without a machine
to run it on, it was hard to hack the kernel, or even know the good
kernel code from the bad with certainty.... Wasn't until my senior
year that the OS course switched over from writing an OS for a
TOPS-20-like machine emulated on a TOPS-20 machine to writing modules
to replace bits of SunOS with your own code...

It was also part of an evolving notion of "OPEN". The SunOS systems
were Open. Totally Open. All the protocols they used were documented
and others could write implementations to them. And there was even a
sample implementation for things like NFS. For the day, that was super
open. Try it with VMS, which had some of the protocols documented and
some of those you could implement w/o running afoul of DEC's (claimed)
IP of various flavors... Sure, it isn't as Open as today, but it was
the first steps down that path...

So Unix has always been an open system. It's just that it's help drive
the notion of Open, including motivating people to work on Linux while
the last bits of it were being freed up and the inevitable legal
hassles that caused.... Of course, various commercial flavors
complicated this picture significantly....  But various commercial
Linux vendors don't really release their sources today, so it can be
hard to get all the bits you need, especially in BSP land.... So Linux
is open, but only mostly open since compliance isn't universal....

Warner

[TUHS] System Economics (was is Linux "officially branded UNIX")

[Dave Horsfall]

On Fri, 17 Mar 2017, Jason Stevens wrote:

> Well $999 would get you source [for BSD/OS]..

Especially if my boss (a very small company) was paying for it...

-- 
Dave Horsfall DTM (VK2KFU)  "Those who don't understand security will suffer."

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Doug McIlroy]

> Many of the gnu tools started life as BSD code that was hacked on and
> rebranded with the GPL.

I have seen Gnu code likewise adopted from AT&T.

Doug

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Doug McIlroy]

> Many of the gnu tools started life as BSD code that was hacked on and
> rebranded with the GPL.

A small amount of code was likewise adopted from AT&T.

Doug

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Nick Downing]

Is this really true, can you give specific examples? AFAIK the GPL cannot
be applied retrospectively except by the BSD- or commercial licensor,
perhaps you could GPL your changes but I am not quite sure how this would
work unless your release was in the form of a patch.
cheers, Nick

On Mar 19, 2017 12:07 AM, "Doug McIlroy" <doug at cs.dartmouth.edu> wrote:

> > Many of the gnu tools started life as BSD code that was hacked on and
> > rebranded with the GPL.
>
> A small amount of code was likewise adopted from AT&T.
>
> Doug
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20170319/dd46c3ae/attachment.html>

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Warner Losh]

You're right. The GPL can't be applied in this way. However, there
were a few attempts (accidental it was claimed) to do this back in the
day, mostly by cutting and pasting bits out of NET2 for this or that
GPL thing. I don't recall the specifics, since it was fixed like
25-odd years ago. Accidental, as claimed, or sneaky, the incidents
(and talk of the incident) left a bad taste in people's mouths. A
couple of times the code in question passed from one person to the
next until the knowledge of the original copying was lost until
discovered by someone who was familiar with the original sources and
did a comparison. The reactions and the personalities didn't help to
smooth over the ruffled feathers either.

To be fair, it was a different time. The knowledge of what was and
wasn't permissible simply isn't at all what it is today. For many
people, it tended to fall into "OK to copy" and "NOT OK to copy". The
nuances of license compliance did not have the benefits of the last
two and a half decades of public education. While some people knew and
respected, it wasn't as universal as it is today. So it was natural
that people would just copy and not attribute. It didn't take too many
incidents of that happening for the word to spread it wasn't cool and
that just because you could copy an entire file w/o a problem doesn't
mean you could cut a dozen routines out of it and paste it into your
own work. That's why any projects that started out as a copy from BSD
(or worse AT&T) were thoroughly reworked to expunge that taint and you
don't hear about it today. It stopped being more than an incidental
problem in the mid 90's. And it wasn't just BSD->GPL either, again to
be fair, the same ignorance allowed code to flow the other way a time
or two...

Warner

On Sat, Mar 18, 2017 at 7:27 AM, Nick Downing <downing.nick at gmail.com> wrote:
> Is this really true, can you give specific examples? AFAIK the GPL cannot be
> applied retrospectively except by the BSD- or commercial licensor, perhaps
> you could GPL your changes but I am not quite sure how this would work
> unless your release was in the form of a patch.
> cheers, Nick
>
> On Mar 19, 2017 12:07 AM, "Doug McIlroy" <doug at cs.dartmouth.edu> wrote:
>>
>> > Many of the gnu tools started life as BSD code that was hacked on and
>> > rebranded with the GPL.
>>
>> A small amount of code was likewise adopted from AT&T.
>>
>> Doug

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Doug McIlroy]

Nick asked for an exmple of AT&T code in Gnu.
Warner explained a spectrum of ways and degrees of innocence
by which that might happen.

The example I have in mind is "calendar" from v7, whose very
idiosyncratic implementation appeared in Gnu with only
cosmetic changes. It has been modified since by discarding
archaic efficiency hacks, but still uses the same quirky
basic method.

Conceivably Gnu's implementation was done only after v7
code was made public. But in any event, it has been
distributed without attribution.

Doug

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Random832]

On Sat, Mar 18, 2017, at 12:25, Doug McIlroy wrote:
> Nick asked for an exmple of AT&T code in Gnu.
> Warner explained a spectrum of ways and degrees of innocence
> by which that might happen.
> 
> The example I have in mind is "calendar" from v7, whose very
> idiosyncratic implementation appeared in Gnu with only
> cosmetic changes. It has been modified since by discarding
> archaic efficiency hacks, but still uses the same quirky
> basic method.

The "calendar" available my Linux machine is from the "bsdmainutils"
package (which has never been put under the GPL, incidentally) which is
stuff copied with attribution from FreeBSD/NetBSD/OpenBSD, and
calendar.c has a UCB copyright/license and OpenBSD RCS ID at the top of
the file. (In the TUHS archive, this implementation first appears in
4.3BSD-Reno, with earlier versions having code clearly derived from V7).

The earliest version (from Debian 1.1) that I can find has a UCB
copyright dated 1993 and SCCS ID 8.3 3/25/94, not much removed from
4.4BSD in the archive (which has SCCS ID 8.1 6/6/93) - my guess is that
the actual source is 4.4BSD-Lite, which is mentioned in the bsdmainutils
README.

In the converted-to-git CSRG archive, the implementation first appears
in 1989 by "bostic" (Keith Bostic, presumably)
https://github.com/weiss/original-bsd/commit/46857f6fe723eff85f22986beb78063f05b60f78
with the change note "redone from scratch as a C program to fix cpp
security problem" - this is the first version to have a UCB copyright
notice. [Using cpp seems to have itself been a BSD innovation circa
4.1cBSD]

V7 calendar consists of a C program that outputs a set of regexes, and a
shell script that runs egrep. Unless by "same quirky basic method" you
mean the fact that it reads events from a text file at all, I'm not sure
what you're referring to... but that's functionality rather than
implementation.

There's a lot of code in a handful of not-part-of-GNU-proper core
utility packages used in Linux distributions - bsdmainutils, bsdgames,
bsdutils, and util-linux [only the last of which has the GPL] - which
come from some BSD or another and mostly have intact UCB copyright
statements and licenses at the top. If any of these are improperly
attributed, it's likely that UCB is to blame.

> Conceivably Gnu's implementation was done only after v7
> code was made public. But in any event, it has been
> distributed without attribution.
> 
> Doug

[TUHS] GNU vs BSD before the lawsuit and before Linux

[Warner Losh]

On Sat, Mar 18, 2017 at 11:45 AM, Random832 <random832 at fastmail.com> wrote:
> There's a lot of code in a handful of not-part-of-GNU-proper core
> utility packages used in Linux distributions - bsdmainutils, bsdgames,
> bsdutils, and util-linux [only the last of which has the GPL] - which
> come from some BSD or another and mostly have intact UCB copyright
> statements and licenses at the top. If any of these are improperly
> attributed, it's likely that UCB is to blame.

Anything in 4.4-lite was specifically blessed by USL as non-infringing
as part of the settlement of that suit...

But the bsd* packages aren't what's being talked about here. Those
generally came about later as these BSD programs ported to Linux and
resistance to non-gpl'd code in distributions waned.

Warner