[TUHS] Is there a good, even definitive, list of reimplementations of the Unix kernel? What would good cut-off criteria be?

[steve jenkin via TUHS]

Below is a slightly edited & cleaned up summary of responses.
Addresses sanitised.

I'll produce a more compact list, probably tomorrow.

My apologies for throwing UTF-8 (smart quotes, smart dashes)
at the list before. Have tried to defeat Apple Mail here.

> On 19 Apr 2026, at 10:32, sjenkin at canb.auug.org.au wrote:
> 
> Original message
> <https://www.tuhs.org/pipermail/tuhs/2026-April/033527.html>



From: John Cowan
Date: 19 April 2026 at 20:05:09 AEST

On Sun, Apr 19, 2026 at 5:110x202FAM Arnold Robbins via TUHS  wrote:


| There was Tunis (I think) done at U. Toronto in Pascal or Modula or something.

Concurrent Euclid, which is also part of the Pascal family.

| IIRC both of those ran on the PDP-11.
| I had a book on Tunis but am not sure if I still have it.

Online at https://archive.org/details/concurrenteuclid00holt.


From: Arnold Robbins via TUHS
Date: 19 April 2026 at 19:10:54 AEST

There was Tunis (I think) done at U. Toronto in Pascal or Modula or something.

There was Idris (I think was the name) done by PJ Plauger.

IIRC both of those ran on the PDP-11. I had a book on Tunis but am not sure if I still have it.

Arnold

Andy Kosela via TUHS wrote:

| What about Coherent?
| Looks to me like a clean reimplementation of UNIX.
| It was quite popular on x86 in the 80s.
| I liked its minimalistic approach compared to other UNIX clones at the time.

| --Andy


From: Andy Kosela
Date: 19 April 2026 at 17:31:59 AEST

What about Coherent?
Looks to me like a clean reimplementation of UNIX.
It was quite popular on x86 in the 80s.
I liked its minimalistic approach compared to other UNIX clones at the time.

--Andy


From: Bakul Shah
Date: 19 April 2026 at 15:16:24 AEST

On Apr 18, 2026, at 8:490x202FPM, Charles H. Sauer wrote:

On Apr 18, 2026, at 9:520x202FPM, Bakul Shah via TUHS wrote:

| I believe Sol later merged into Chorus was a reimplementation in Pascal. I think this was a microkernel based system...
| I believe Locus distributed OS designed at UCLA was also Unix compatible. I guess you can add Xinu as well as Amoeba to the list.

Thanks. I have the Locus book but couldn't find it.
If they started with BSD code, they must've had to do major surgery to achieve a distributed system!
Except for bootstrapping & drivers probably easier to start from scratch....

I guess KeyNIX (atop KeyKOS) should be added as well:
    http://cap-lore.com/CapTheory/KK/UnixOnMicroKernel/


From: "Charles H. Sauer via TUHS"
Date: 19 April 2026 at 13:49:09 AEST

On Apr 18, 2026, at 9:520x202FPM, Bakul Shah via TUHS wrote:

| I believe Sol later merged into Chorus was a reimplementation in Pascal. I think this was a microkernel based system...
| I believe Locus distributed OS designed at UCLA was also Unix compatible. I guess you can add Xinu as well as Amoeba to the list.

I intentionally never looked at Locus source,
but had much interaction with Gerry Popek and Bruce Walker while I was at IBM and they were working with IBM.

I have a memory of Bruce telling me they started with 4.1BSD, but I question that memory.
Just glancing at their book
    (https://www.penguinrandomhouse.com/books/655167/the-locus-distributed-system-architecture-by-gerald-j-popek/),
the only cited Unix reference I see is the 1978 Ritchie/Thompson BSTJ paper.

Section 1.5 Unix Compatibility of their book says
    "For virtually all applications code, the LOCUS system can provide complete compatibility,
     at the object code level, with both Berkeley Unix and System V, ..."

I suspect that the first Locus prototypes were based on BSD earlier than 4.1.

Charlie
--


From: Bakul Shah
Date: 19 April 2026 at 12:52:03 AEST

I believe Sol later merged into Chorus was a reimplementation in Pascal. I think this was a microkernel based system...
I believe Locus distributed OS designed at UCLA was also Unix compatible.
I guess you can add Xinu as well as Amoeba to the list.


From: Peter Yardley
Date: 19 April 2026 at 15:55:58 AEST

Hi,

OSX uses a microkernel, Mach.
Thing is more and more code got moved from user space into the kernel to speed it up.

First OSF1 moved more code into kernel space, then NeXT, then Apple.
But the kernel definitely started out as the Mach micro kernel.

On 19 Apr 2026, at 11:160x202Fam, John Levine via TUHS wrote:

| According to steve jenkin via TUHS:
| | Was "Wind River" and its VxWorks "Unix-like" or something else?
| 
| Something else, I think.  They bought BSDI to get a Unix system.
| 
| | Does GNU count, as their HURD kernel has never attained the attributes necessary for wide use?
| 
| HURD lost its reason for existing once linux worked since it's also GPL.
| I see people poking at HURD, but nothing serious.
| Microkernels are an idea that looks appealing but has never really
| worked because they're slow due to all the process switching.

.1.3.6.1.4.1.8852.4.2
Peter Yardley


From: John Levine via TUHS
Date: 19 April 2026 at 11:16:57 AEST

According to steve jenkin via TUHS:
| Was "Wind River" and its VxWorks "Unix-like" or something else?

Something else, I think.  They bought BSDI to get a Unix system.

| Does GNU count, as their HURD kernel has never attained the attributes necessary for wide use?

HURD lost its reason for existing once linux worked since it's also GPL.
I see people poking at HURD, but nothing serious.

Microkernels are an idea that looks appealing but has never really
worked because they're slow due to all the process switching.


From: Larry McVoy
Date: 19 April 2026 at 10:54:04 AEST

I have heard, but not seen the code,
that the Data General Unix was from scratch and was very good.

People talk about it like I talk about SunOS 4.x
which is by far the best Unix kernel I have ever seen.

-- 
---
Larry McVoy           Retired to fishing

--


From: Theodore Tso via TUHS
Date: 19 April 2026 at 13:40:28 AEST

On Sun, Apr 19, 2026 at 10:32:14AM +1000, steve jenkin via TUHS wrote:

| We know that implementing a V7 kernel API was a masters project for
| Linus, taking ~3 years to crank it out, presumably his sole work as
| a thesis requirement.

If you meant Linus Torvalds, that's not correct.

Linus announced Linux on comp.os.minux on August 26, 1991,
and and in it, he claimed that it had been "brewing since April."

The first snapshot of his Linux, 0.01,
was released on ftp.funet.fi on September 17, 1991.

In 1991, Linux would have been roughly in his third year,
since he started at the University of Helsink in 1988,
but then in 1989 he took a year off to fulfill his mandatory military service,
so he would have started his Sophomore year in the fall of 1990,
so Linux 0.01 which would have full V7 kernel functionality
at the beginning of his Junior year in September, 1992.

Linus did use Linux as the basis of his Masters thesis[1],
but that thesis was submitted on January 31, 1997,
and that was long after Linux had accumulated a large number of contributors,
with Linux 2.0.28 being released in January 14, 1997,
and in 1996 was when Linux 2.0 was released with SMP support,
and when Larry Ewing created the Tux penguin mascot.

[1] https://www.cs.helsinki.fi/u/kutvonen/index_files/linus.pdf

Cheers,

						- Ted

From: Luther Johnson
Date: 19 April 2026 at 10:45:39 AEST

You might consider the xv6 project as a candidate for clean/pedagogical re-implementation of V6 UNIX:

    https://pdos.csail.mit.edu/6.828/2012/xv6.html

--
Steve Jenkin, IT Systems and Design 
0412 786 915 (+61 412 786 915)
PO Box 38, Kippax ACT 2615, AUSTRALIA

mailto:sjenkin at canb.auug.org.au http://members.tip.net.au/~sjenkin