News for hackers. Stuff that matters.
Re:Porting Unix to the 386: a Practical Approach (1991) by DrScump on Wednesday July 19, 2017 @02:03AM (#14802257)
More specifically, 386BSD.

There had already been commercial UNIX System V ports to 80*86 (including the 8088) such as the AT&T 6300 Plus (and, arguably, SCO).

Re:Porting Unix to the 386: a Practical Approach (1991) by yuhong on Wednesday July 19, 2017 @03:23AM (#14802527)
Indeed, one of the first 386 OSes was Xenix. In retrospect, I really wished that DRI added GEM and native 32-bit application support into Concurrent DOS 386. GEM would probably be built into the kernel, with virtual 8086 mode used to run legacy 16-bit GEM apps and developers could write new 32-bit GEM (and even command-line) apps. It would be far more interesting than DR-DOS and probably a good competitor to even OS/2 2.0.
Re:Porting Unix to the 386: a Practical Approach (1991) by flyinghamster on Wednesday July 19, 2017 @11:10AM (#14804843)
Xenix was also ported to a variety of other architectures as well. At my very first programming job, we had a Z8000 running a very early (V7-like) Xenix port. Its performance was not impressive; a 386 could run rings around it. Also, the segmented memory architecture limited processes to 64K instructions and 64K data.

At a later job, we had several projects that we put together on the 386 version of Xenix; one of those is still in service today. It had moved to OpenServer relatively early on, and stayed that way until early this summer; my task was to get it running on a modern Linux. I was very (pleasantly) surprised to discover that the Unify database is still available, even for RHEL7 - and it was possible to directly take the database files from the old system and use them unmodified on the new one.

I wonder what prompted Microsoft to give up on Xenix? AT&T wanting too much money, or perhaps Not-Invented-Here syndrome? Windows could have been an X desktop.

Re:Porting Unix to the 386: a Practical Approach (1991) by gaius on Thursday July 20, 2017 @11:38AM (#14813405)
wonder what prompted Microsoft to give up on Xenix

They reassigned the Xenix devs to OS/2 and sold Xenix to SCO (who were a real company back then).

Re:Porting Unix to the 386: a Practical Approach (1991) by osullivj on Wednesday July 19, 2017 @10:50AM (#14804665)
I ported a large Fortran codebase for mech eng analysis to Xenix from VMS back in the late 80s. We needed more than 640K, so DOS wouldn't do, and this was before Windows. I had to write a device driver for the 68000 based graphics card. That was my first real C coding experience. Fun! I didn't know about BSD at the time, but it looks like Xenix was System V [1].

[1] https://en.wikipedia.org/wiki/Xenix

Re:Porting Unix to the 386: a Practical Approach (1991) by DrScump on Wednesday July 19, 2017 @11:39PM (#14809877)
Did you have a means of using extended or expanded memory?

There were memory libraries like DOS/16M that would allow use of extended memory and 80286+ Protected Mode without a lot of source changes. Informix, Lotus, etc. used it for their Protected Mode product line on DOS starting in the late 1980s.

Re:Porting Unix to the 386: a Practical Approach (1991) by dvfjsdhgfv on Wednesday July 19, 2017 @02:36AM (#14802368)
The authors explicitly said they want to build something more free: "We hope to see those interested in this technology build on it in both commercial and noncommercial ventures." And they did port Unix (Berkeley flavor) to the 386. I remember using it in the nineties, it was a great adventure. And its progeny FreeBSD is arguably the best alternative to Linux, should something go terribly wrong.
Re:Porting Unix to the 386: a Practical Approach (1991) by DrScump on Wednesday July 19, 2017 @03:30AM (#14802541)
I was just clarifying the title, not casting any aspersions.

My first UNIX environment was System V on AT&T's 3B series. We subsequently moved to Sun 3s with BSD. I strongly preferred the latter.