Sure. (Perhaps I used an incorrect term, but "paging" seemed reasonable.) The executing image consisted of 3 parts: 2 8K segments of instructions, and 1 48K segment of data/stack. One of the 8K instruction segments was always resident in core and contained some program code and sort of jump table. As the program executed, it used the jump table to map appropriate 8K sections of code into the other 8K instruction segment. The 48K data segment was always resident in memory. The net effect was a slow pseudo split I&D area with 56K instruction space and 48K data space. For the most part, this was adequate to run system utilities. The link edit scheme was fairly complicated. You first had to build two executables: a standard shared text executable, and a standard split I&D executable. Then you had to run a special program which took those two binaries as input and constructed the final image. If any of your subroutines couldn't fit into 8K, you were out of luck. Fortunately, the C compiler driver had a option to do this silliness automatically. I remember being frustrated by the fact that Microsoft had, for reasons unknown, forgotten to build /usr/bin/sort as one of these special binaries. Instead, it was a conventional split I&D executable. I had to convince a friend of mine whose institution had a source license to let me "borrow" the source to sort so I could build an executable that ran on a PDP 11/23. Robert Tillyard asked "Would SCO->Caldera have copies of this [PDP/11 XENIX]?" I remember asking that same question on this very list several years ago. Warren's answer was essentially "No." If I remember correctly, he said that they really had no archival material at all. I don't remember what happened to the copy I worked with. It probably vanished into a dumpster when the company I worked for went "belly up." If anyone *does* still have a copy, it would be great to see it in the PUPS Archives, since it is probably the "rarest" PDP/11 Unix variant. -- Frank "They haven't got Brains, any of them, only grey fluff that's blown into their heads by mistake, and they don't Think." * Eeyore, "The House at Pooh Corner"

on 3/26/02 2:41 AM, Michael Davidson at MichaelDavidson(a)pacbell.net wrote: Yes! Thank you for reminding me; I had forgotten its name. Too bad. It would be nice if a tape of XENIX/11 surfaced. I'll look in my attic again, but I doubt I have anything from that era. -- Frank "Plez cnoke if an rnsr is not reqid." Sign on Owl's Door, "Winnie the Pooh"

In article by Robert Tillyard: SCO/Caldera only have what we have given them from the Unix Archive. That's pretty ironical 8-) I assume that any tapes of Xenix we found would be (c) HCR or Microsoft. I don't even want to contemplate getting the release permissions :-( Warren

I'm getting really confused with all these companies. If I understand correctly... AT&T/Western Electric sold UNIX rights to Microsoft. Microsoft had HCR develop XENIX from V7. SCO licensed XENIX from Microsoft. SCO then subsubsublicensed XENIX to various vendors. Please correct me. I must be wrong. (What happened to our MERT discussion anyway? :-) Long & winding PS.: I read this really cute book about Linux at my local library some time ago. It discussed UNIX, Linux, their relation, and the current state of affairs when it was written -- in 1994. The book started out with an etymology of XENIX, which would have been derived from Dutch "'k Zie niks," meaning "I don't see a thing" -- the first thing Dutch users uttered when XENIX booted. Last year, I talked to a fellow member of the HCC (Hobby Computer Club) UNIX gg (gebruikersgroep, user group) who remembered his worst experience with UNIX -- having to use XENIX. He was still shocked by its Microsoftian performance.

Thus spake Michael Davidson <MichaelDavidson(a)pacbell.net>: thanks for trying. it helped! I meant licensed. sorry. Tandy, right? this brings up memories of my first experiences in computing... being +-6 years old and impressed by my cousin's 286/20 (we had a /12)

On 2002-Mar-28 11:58:08 +1030, Greg 'groggy' Lehey <grog(a)lemis.com> wrote: I recall using XENIX/286 in the mid 80's. We were porting some software from Micro/RSX on a PDP-11/03. The AT-class box (an ITT XTRA/XL) was significantly faster. And later, having 8 developers on one 8MHz 80286 was, err, interesting. I also remember having it panic when it tried to fork a large (~2MB) process. And unbundled. I don't recall too many problems with the C compiler, but we were porting Pascal - and the Pascal compiler was atrocious. The object code (and whether a piece of code would compile at all) seemed to depend on whim of the compiler. Obviously whoever wrote it hadn't grasped the concepts of initialising variables, making arrays large enough to fit their contents or only using a single piece of memory to store one thing at a time :-). I don't think you could get any of the GNU tools to run on a 286 so I was mostly limited to the SCO tools. I recall doctoring MicroEmacs so that it was all small model but allocated the buffers in "far" memory - that gave me the ability to edit large files with acceptable performance. Peter

On Saturday, 30 March 2002 at 12:03:28 +0100, Lars Buitinck wrote: No, Siemens and Nixdorf merged in the early 90s. Interesting derivation, but Sinix has been around since before Siemens and Nixdorf merged. Greg -- Finger grog(a)lemis.com for PGP public key See complete headers for address and phone numbers