Hi Clem,
At 2024-05-25T12:21:17-0400, Clem Cole wrote:
> On Sat, May 25, 2024 at 12:13 PM G. Branden Robinson <
> g.branden.robinson@gmail.com> wrote:
>
> > That does complicate my simplistic story. Ing70 was, then, as you noted
> > in a previous mail, an 11/70, but it _wasn't_ running Version 7 Unix,
> > but rather something with various bits of BSD (also in active
> > development, I reckon).
> >
> Mumble -- the kernel and 90% of the userspace on Ing70 was V7 -- it was
> very similar to Teklabs which I ran.
Yes, sorry, I was hasty and sloppy. I should have qualified that
"Version 7 Unix" with "pure". Though I wonder if anyone ran "pure"
distributions of anything by today's standards, with our flatpaks and VM
images and containers and distributions and Linux kernel "taint" flags.
And, blessed be, our reproducible builds. So there is such a thing as
progress.
> The point is it was a 16 bits system, the Johnson C compiler with some
> fixes from the greater USENIX community including UCB.
> There was >>no port<< needed.
>
> This was its native tongue.
Okay. My crystal ball shows wordsmithing in my future.
> It was >>included<< in later BSD released which is how people came to
> know it because 4.XBSD was became much more widely used than V7+2BSD.
Acknowledged.
> The 2.9 work of Keith at al, started because the UCB Math Dept could
> not afford a VAX. DEC had released the v7m code to support
> overlays, so slowly changed from the VAX made it back into the V7
> based kernel - which took a new life.
Ah, I'd never heard the actual origin story of later 2BSD's reason for
parallel development. Thanks!
The 2.8 kernel from the 2.83 archive is a V7 with a bunch of hacks / features #ifdef'd into the tree with a primitive config thing to cons up the #defines. This is still largely present in 2.9, but with less rigid adherence for bug fixes. It's very clear that for the kernel this was followed. I've not studied userland to comment on that but i think not.
It also explains why the release notes kept saying it was the last release starting iirc with 2.8...
Warner
Back when I was first learning Unix, a mere 30 years ago, I asked a
local guru why the kernel image was called "vmunix" instead of just
plain "unix". I got a correct answer, but then asked why you'd keep
calling it "vmunix" when no non-VM Unix was even available for the
platform. Historical inertia and the long shadow of the work that
became 4BSD. (Linus's decision to name his kernel's image "vmlinux" [or
"vmlinuz" for those remember having those lulz] when in its case no
non-VM version had ever existed anywhere, nor even been desired or
conceived, struck me as an excess of continuity.)
Unix geeks are conservative about the weirdest things.
Regards,
Branden