[TUHS] Open sourcing SunOS?

[TUHS] Open sourcing SunOS?

[Dan Cross]

Has anyone tried talking to anyone at Oracle about possibly getting
the SunOS code released under an open source license? There can't be
any commercial value left in it.

        - Dan C.

[TUHS] Re: Open sourcing SunOS?

[Robert Clausecker]

That's what OpenSolaris was.  The bits that weren't open sourced
were the bits they weren't allowed to for licensing reasons, such
as the C compiler.

Yours,
Robert Clausecker

Am Wed, Feb 22, 2023 at 10:49:50AM -0500 schrieb Dan Cross:
> Has anyone tried talking to anyone at Oracle about possibly getting
> the SunOS code released under an open source license? There can't be
> any commercial value left in it.
> 
>         - Dan C.

-- 
()  ascii ribbon campaign - for an 8-bit clean world 
/\  - against html email  - against proprietary attachments

[TUHS] Re: Open sourcing SunOS?

[Dan Cross]

On Wed, Feb 22, 2023 at 11:23 AM Robert Clausecker <fuz@fuz.su> wrote:
> That's what OpenSolaris was.  The bits that weren't open sourced
> were the bits they weren't allowed to for licensing reasons, such
> as the C compiler.

No. OpenSolaris was the SVR4-based Solaris 2.  I'm referring to the
BSD-based SunOS (e.g., 4.1.4, 4.1.3U1, etc).

        - Dan C.

[TUHS] Re: Open sourcing SunOS?

[Dave Horsfall]

On Wed, 22 Feb 2023, Dan Cross wrote:

> No. OpenSolaris was the SVR4-based Solaris 2.  I'm referring to the 
> BSD-based SunOS (e.g., 4.1.4, 4.1.3U1, etc).

I'd love to see the magic behind 4.1.4...  Best OS I'd ever used.

-- Dave

[TUHS] Re: Open sourcing SunOS?

[Warner Losh]

[-- Attachment #1: Type: text/plain, Size: 2903 bytes --]

On Wed, Feb 22, 2023, 8:50 AM Dan Cross <crossd@gmail.com> wrote:

> Has anyone tried talking to anyone at Oracle about possibly getting
> the SunOS code released under an open source license? There can't be
> any commercial value left in it.
>

SunOS 4 has a lot of encumbered code in it, especially for i386 drivers.
Bits of the
network stack as well. It was hopeless to try to open source. There was a
lot of bits
and pieces that Sun had done with contracts that were, at best, ambiguous
for
what to do should they want to open source it. At least that's the story
I've heard
from people at Sun, both engineers and management.

I know when Solbourne did their OS/MP 4.0 and 4.1 SunOS MP system, there
was a lot of back and forth between Solbourne's and Sun's lawyers to get all
the pieces needed to build the system. For example, initially SunView was
not
included in the offering, so my group was formed to write our clone for it
(This
would become OI (Object Interface) and uib (User Interface Builder) that we
did in C++ to allow it to present in either Motif or OpenLook).

I also know that Sun tried to donate their VM system to Berkeley btween
BSD4.3
and BSD4.4. Had the support of Scott McNeely and was almost a done deal.
However,
the lawyers said that the company would need to take a 'write down' loss on
the
donation, which would likely tank the stock price of Sun, so it was nixed.
So instead,
Berkeley did the next best thing (really the only available thing) and went
with the
MachVM.

One can find the sources to 4.1.4 and 4.1.3 online (though not the earlier
4.1.1 or
4.0.x which have the i386 port in it). The newer sources do have i386 bits,
but from
my cursory investigation, it's mostly remnants of the support. Of course,
if one were
to modify this, and try to distribute it, you'd likely be asking for
trouble. At least two
of the abandonware sites have copies. There's also something purporting to
the the
SCCS files for the system, but alas, they are not actually there.

Having talked to the VPs that got OpenSolaris released about doing SunOS
(after
they'd left Sun, I worked for them), it was clear that OpenSolaris was hard
enough.
SunOS 4 was too old by even the point to get all the successors in interest
to
agree to modifications of the old contracts to allow source distribution
without
significant restrictions to agree to make enough of the system available
for it to
be interesting to people.

Oracle has pulled back from OpenSolaris, fired all the Solaris engineers
and has
all that on basically life support. I doubt they'd want to do anything for
even that
modern code base, let alone something that's ancient history, would be hard
to find,
would be difficult to find the old paperwork for it to clear legal, etc.

So, I'm not at all optimistic this could ever happen prior to the
expiration of the
copyrights of the original code.

Warner

        - Dan C.
>

[-- Attachment #2: Type: text/html, Size: 3998 bytes --]

[TUHS] Re: Open sourcing SunOS?

[Larry McVoy]

On Wed, Feb 22, 2023 at 01:04:37PM -0700, Warner Losh wrote:
> On Wed, Feb 22, 2023, 8:50 AM Dan Cross <crossd@gmail.com> wrote:
> 
> > Has anyone tried talking to anyone at Oracle about possibly getting
> > the SunOS code released under an open source license? There can't be
> > any commercial value left in it.
> >
> 
> SunOS 4 has a lot of encumbered code in it, especially for i386 drivers.

There is SunOS as in everything shipped, kernel and userspace, and there
is the kernel.  So far as I remember, the i386 stuff was never integrated
into the source tree that Sun shipped from.  There was the roadrunner
stuff but I don't think that ever made it in to the official tree.  If
it did, nobody paid attention to it.  All people cared about at the 
time as SPARC and I don't think there was any outsourced hacking for
SPARC, that was all in house.

The networking stack in SunOS 4.x was BSD derived.  You might be thinking
of Solaris, that took the Lachman STREAMS stack but that was 5.x, not
4.x.

As the only guy, that I'm aware of, who took all the encumbered stuff
out of the kernel, put back the BSD tty drivers and a few other small
things that resulted in a kernel that we could freely open source,
I beg to differ with:

> Bits of the
> network stack as well. It was hopeless to try to open source. There was a
> lot of bits
> and pieces that Sun had done with contracts that were, at best, ambiguous
> for
> what to do should they want to open source it. 

[TUHS] Re: Open sourcing SunOS?

[Brad Spencer]

Larry McVoy <lm@mcvoy.com> writes:

> On Wed, Feb 22, 2023 at 01:04:37PM -0700, Warner Losh wrote:
>> On Wed, Feb 22, 2023, 8:50 AM Dan Cross <crossd@gmail.com> wrote:
>> 
>> > Has anyone tried talking to anyone at Oracle about possibly getting
>> > the SunOS code released under an open source license? There can't be
>> > any commercial value left in it.
>> >
>> 
>> SunOS 4 has a lot of encumbered code in it, especially for i386 drivers.
>
> There is SunOS as in everything shipped, kernel and userspace, and there
> is the kernel.  So far as I remember, the i386 stuff was never integrated
> into the source tree that Sun shipped from.  There was the roadrunner
> stuff but I don't think that ever made it in to the official tree.  If
> it did, nobody paid attention to it.  All people cared about at the 
> time as SPARC and I don't think there was any outsourced hacking for
> SPARC, that was all in house.
>
> The networking stack in SunOS 4.x was BSD derived.  You might be thinking
> of Solaris, that took the Lachman STREAMS stack but that was 5.x, not
> 4.x.
>
> As the only guy, that I'm aware of, who took all the encumbered stuff
> out of the kernel, put back the BSD tty drivers and a few other small
> things that resulted in a kernel that we could freely open source,
> I beg to differ with:
>
>> Bits of the
>> network stack as well. It was hopeless to try to open source. There was a
>> lot of bits
>> and pieces that Sun had done with contracts that were, at best, ambiguous
>> for
>> what to do should they want to open source it. 

There may have been other parallel efforts in one form or the other.  I
know that there existed a patch to SunOS 4.1.3 that updated the network
stack to a newer version from Berkeley that gained a couple of new
features over the one Sun delivered.  The patch was source code, but it
was possible with this patch to apply it to a binary copy of SunOS and
recompile / compile a new kernel that had the new stack in it, that is,
you didn't need the full source to the kernel.  It was a long time ago
and I don't remember the details exactly, but I did use it on a file /
build / NIS server we had in the department at AT&T/Lucent where I was
at.  My point mostly being that hacking on SunOS 4.x appears to have
happened here and there.  It certainly would have been nice to have a
open source SunOS 4.x around although the userland may have presented
its own trouble.





-- 
Brad Spencer - brad@anduin.eldar.org - KC8VKS - http://anduin.eldar.org

[TUHS] Re: Open sourcing SunOS?

[Warner Losh]

[-- Attachment #1: Type: text/plain, Size: 2633 bytes --]

On Wed, Feb 22, 2023 at 1:12 PM Larry McVoy <lm@mcvoy.com> wrote:

> On Wed, Feb 22, 2023 at 01:04:37PM -0700, Warner Losh wrote:
> > On Wed, Feb 22, 2023, 8:50 AM Dan Cross <crossd@gmail.com> wrote:
> >
> > > Has anyone tried talking to anyone at Oracle about possibly getting
> > > the SunOS code released under an open source license? There can't be
> > > any commercial value left in it.
> > >
> >
> > SunOS 4 has a lot of encumbered code in it, especially for i386 drivers.
>
> There is SunOS as in everything shipped, kernel and userspace, and there
> is the kernel.  So far as I remember, the i386 stuff was never integrated
> into the source tree that Sun shipped from.  There was the roadrunner
> stuff but I don't think that ever made it in to the official tree.  If
> it did, nobody paid attention to it.  All people cared about at the
> time as SPARC and I don't think there was any outsourced hacking for
> SPARC, that was all in house.
>

I know there's i386 source kits for the Roadrunner, and that's really the
only thing that SunOS ever supported, at least that was sold. I'd been told
that it had been poorly integrated by people that had worked on it here in
Boulder, so I believe that.

In the 4.1.3 sources there's a number of i386 ifdefs, and at least a math
library
for i386. But no kernel bits, except for a bunch of ifdefs for i386 that
I've not
looked at closely. Maybe this is what is meant by 'poorly integrated' :)


> The networking stack in SunOS 4.x was BSD derived.  You might be thinking
> of Solaris, that took the Lachman STREAMS stack but that was 5.x, not
> 4.x.
>

I was thinking of the streams stuff that's in 4.x BSD. There's AT&T
copyrights
on it. There's also, strangely, rfs sources included with some of the stuff
one
can find online. But it looks to be imported nearly verbatim from System V
of some flavor with very few edits, judging by the 1.1 versions in many of
the
files.


> As the only guy, that I'm aware of, who took all the encumbered stuff
> out of the kernel, put back the BSD tty drivers and a few other small
> things that resulted in a kernel that we could freely open source,
> I beg to differ with:
>
> > Bits of the
> > network stack as well. It was hopeless to try to open source. There was a
> > lot of bits
> > and pieces that Sun had done with contracts that were, at best, ambiguous
> > for
> > what to do should they want to open source it.
>

I'm just reporting what my VP told me...  Grepping through the source I can
find online, the evidence is closer to what you say than what Glen told me.
Either he or I must have confused Solaris 2 with SunOS 4.

Warner

[-- Attachment #2: Type: text/html, Size: 3739 bytes --]

[TUHS] Re: Open sourcing SunOS?

[Larry McVoy]

On Wed, Feb 22, 2023 at 01:46:53PM -0700, Warner Losh wrote:
> On Wed, Feb 22, 2023 at 1:12 PM Larry McVoy <lm@mcvoy.com> wrote:
> > The networking stack in SunOS 4.x was BSD derived.  You might be thinking
> > of Solaris, that took the Lachman STREAMS stack but that was 5.x, not
> > 4.x.
> >
> 
> I was thinking of the streams stuff that's in 4.x BSD. There's AT&T
> copyrights
> on it. There's also, strangely, rfs sources included with some of the stuff
> one
> can find online. But it looks to be imported nearly verbatim from System V
> of some flavor with very few edits, judging by the 1.1 versions in many of
> the
> files.

I ripped out the STREAMS stuff (it was STREAMS, not dmr's streams).
And I ripped out RFS.  There was more than 1.1 versions in some version
of SunOS, my former office mate, Howard Chartok, did a pile of work on
RFS.

> > As the only guy, that I'm aware of, who took all the encumbered stuff
> > out of the kernel, put back the BSD tty drivers and a few other small
> > things that resulted in a kernel that we could freely open source,
> > I beg to differ with:
> >
> > > Bits of the
> > > network stack as well. It was hopeless to try to open source. There was a
> > > lot of bits
> > > and pieces that Sun had done with contracts that were, at best, ambiguous
> > > for
> > > what to do should they want to open source it.
> 
> I'm just reporting what my VP told me...  Grepping through the source I can
> find online, the evidence is closer to what you say than what Glen told me.
> Either he or I must have confused Solaris 2 with SunOS 4.

Like I said, in 1992 or 1993, I had a BSD licensed SunOS 4.1.something, 
I think 4.1.3, kernel.  No STREAMS, no RFS, no STREAMS based tty drivers,
it was what lots of people called SunOS: "A bugfixed and improved BSD".

I shopped it around inside Sun and there were plenty of people who wanted
a reason to say it wasn't open source ready and they couldn't find one.
-- 
---
Larry McVoy           Retired to fishing          http://www.mcvoy.com/lm/boat

[TUHS] Re: Open sourcing SunOS?

[Rob Gingell]

On 2/22/23 12:04 PM, Warner Losh wrote:
> SunOS 4 has a lot of encumbered code in it, ...

SunOS had a complicated set of license encumbrances. I can't claim to 
have fully understood them even at the time. In the mid-1980s, the 
notion of "open sourcing" as we understand it today wasn't a goal or 
even considered and so never entered into the terms of the technologies 
Sun used. These might have been overcome with effort but in the context 
of the time it didn't seem important.

Things Sun licensed widely, like source kits for NFS/VFS, were things we 
routinely and repeatedly sanitized as they evolved.

As part of the SPARC partners program SunOS was licensed to people 
building SPARC-based products but as Warner notes in the Solbourne 
experience, especially early on, there wasn't a "product" so much as a 
"process" that disseminated it.

> I also know that Sun tried to donate their VM system to Berkeley btween 
> BSD4.3
> and BSD4.4. 

It would be more correct to say that Sun was willing to donate the VM 
system back to Berkeley, but my recollection is that CSRG planned to get 
to that functionality through a different implementation path and didn't 
want it.

We expected "everyone" to eventually get the VM system, as it was in 
SVR4 before SunOS 4.0 even shipped, and so "everyone" (even the people 
who otherwise offered BSD systems) would have access to it, just like we 
did. (Of course that notion of "everyone" is pretty limited but at the 
time it was just The Way It Was.)

We did donate all the shared library work to Berkeley, probably the 
closest to what we'd now call "open source" that Sun did in that era. At 
the time, Berkeley didn't plan on migrating off of the a.out object file 
format and so it was useful to them to have the a.out-based implementation.

> Had the support of Scott McNeely and was almost a done deal. However > the lawyers said that the company would need to take a 'write down' loss
> on the
> donation, which would likely tank the stock price of Sun, so it was 
> nixed. 
While I am confident he would have supported it, I doubt Scott ever knew 
about the donations and discussions. The organization-chart-local VP 
signed off on it and I had a 10 minute phone conversation (no part of 
which involved balance sheets or stock values) with the lawyer who wrote 
the letter of transmittal that accompanied the code.

[TUHS] Re: Open sourcing SunOS?

[Warner Losh]

[-- Attachment #1: Type: text/plain, Size: 3039 bytes --]

On Wed, Feb 22, 2023 at 5:12 PM Rob Gingell <gingell@computer.org> wrote:

> On 2/22/23 12:04 PM, Warner Losh wrote:
> > SunOS 4 has a lot of encumbered code in it, ...
>
> SunOS had a complicated set of license encumbrances. I can't claim to
> have fully understood them even at the time. In the mid-1980s, the
> notion of "open sourcing" as we understand it today wasn't a goal or
> even considered and so never entered into the terms of the technologies
> Sun used. These might have been overcome with effort but in the context
> of the time it didn't seem important.
>
> Things Sun licensed widely, like source kits for NFS/VFS, were things we
> routinely and repeatedly sanitized as they evolved.
>
> As part of the SPARC partners program SunOS was licensed to people
> building SPARC-based products but as Warner notes in the Solbourne
> experience, especially early on, there wasn't a "product" so much as a
> "process" that disseminated it.
>

Process is a good word here :).


> > I also know that Sun tried to donate their VM system to Berkeley btween
> > BSD4.3
> > and BSD4.4.
>
> It would be more correct to say that Sun was willing to donate the VM
> system back to Berkeley, but my recollection is that CSRG planned to get
> to that functionality through a different implementation path and didn't
> want it.
>

I got my story from Kirk McKusick who  was very clear that it was going
to happen, but didn't due to some hold up at the highest levels. It was only
then that the decided to go with Mach after they couldn't get it...

We expected "everyone" to eventually get the VM system, as it was in
> SVR4 before SunOS 4.0 even shipped, and so "everyone" (even the people
> who otherwise offered BSD systems) would have access to it, just like we
> did. (Of course that notion of "everyone" is pretty limited but at the
> time it was just The Way It Was.)
>

Yea, all proper licensees :).


> We did donate all the shared library work to Berkeley, probably the
> closest to what we'd now call "open source" that Sun did in that era. At
> the time, Berkeley didn't plan on migrating off of the a.out object file
> format and so it was useful to them to have the a.out-based implementation.
>

Yea, BSD didn't migrate off of a.out until the fission into FreeBSD and
NetBSD
since they did things slightly differently...


> > Had the support of Scott McNeely and was almost a done deal. However >
> the lawyers said that the company would need to take a 'write down' loss
> > on the
> > donation, which would likely tank the stock price of Sun, so it was
> > nixed.
> While I am confident he would have supported it, I doubt Scott ever knew
> about the donations and discussions. The organization-chart-local VP
> signed off on it and I had a 10 minute phone conversation (no part of
> which involved balance sheets or stock values) with the lawyer who wrote
> the letter of transmittal that accompanied the code.
>

 I don't doubt that. I certainly wasn't there. I've heard the story from
Kirk several times though...

Warner

[-- Attachment #2: Type: text/html, Size: 4430 bytes --]

[TUHS] Re: Open sourcing SunOS?

[Rob Gingell]

On 2/22/23 6:14 PM, Warner Losh wrote:
>   I don't doubt that. I certainly wasn't there. I've heard the story 
> from Kirk several times though...

On the "it would tank the stock" story, it's certainly a much better 
story than the boring one I have. I'm thinking legs are getting pulled 
somewhere on that. Though were it true it would have the pleasant 
property of making our engineering efforts be reflected in the business 
in a much more inflated way, which, as several people have pointed out 
in the past isn't how it usually works.

On 2/22/23 6:14 PM, Warner Losh wrote:
> I got my story from Kirk McKusick who  was very clear that it was going
> to happen, but didn't due to some hold up at the highest levels. 
While I'm confident the "tank the stock" story isn't true, that Kirk 
remembers there being a hold up makes me concerned that I'm not 
remembering everything. Perhaps my memories are too rosy based on what 
we were able to do, such that I'm overlooking some consideration that 
would have made the VM system different. Kirk (and Mike Karels) had been 
involved in the work early on so it benefited from their help and they 
were certainly able to evaluate it in the context of what they wanted to 
do with BSD. Guess I need to dig and reconstruct the story better...