Caldera releases original unices under BSD license

q[alex] writes "Caldera International has done a very good thing. They have released the "Ancient" Unices they inherited when they purchased SCO under a "BSD-style" license. The license is available here, instructions on finding the source are here. Caldera (and before that SCO) had required people to obtain a free (as in beer) but somewhat restrictive license in order to get these old sources. The new BSD-style licensing only applies to the 16-bit PDI-11 versions and some of the early 32-bit releases (excluding System III and System V), but it's still very cool."

Comments

Thanks -- and not just to Caldera.

[Anonymous Coward]

I think it's wonderful that Caldera decided, pretty much without even being directly asked, to free the sources to historical Unix.

It's noteworthy that they decided to do this just a few days after, finally, the successful end of Perry Metzger's long campaign to free the historical Unix *documentation* (perhaps more useful, these days). It sure seems to me that once they seriously considered Perry's request they must have realized the PR benefits they could reap by freeing the source code, too.

Thanks, Caldera -- and if you too are grateful to Caldera, you should probably be grateful to Perry as well.

Re:I've been using this same software for a year n

[Saint Aardvark]

Absolutely. Someone's already put a link to Amazon for the book, so go check it out. It's amazing how damned interesting it is to read it, to see how things like task switching and process creation are actually done. (Keep in mind I came into Unix as a hobby, so I haven't taken any of this in CS courses.)

Someone was asking up above what the point was. Aside from learning from reading yet more code, this is as close as we can get to original Unix. If I was given the chance to check out the original manuscript for, say, the Revelation of St. John (I'm atheist, but religion fascinates me), I'd jump at the chance to see what changes have been made between the original and what we've got now. You'd learn an awful lot about how things have changed -- not just the book itself but everything else. I think that would apply with Unix just as much.

Couple things: buckrogers, I downloaded the source for the PDP-11 version a year or so ago. The original tarball has been lost in the mists of hd upgrades, but I do remember not being able to correlate Lions' code sections with the source I downloaded. Has your experience been any different?

And another thing: I'm sure I went picked up the code -- just the code -- for the PDP-11 version, way back when. Now, though, I can't seem to find it on their site. I thought I checked through the directories pretty thoroughly -- can anyone tell me where it is, or provide a link to their own copy?

Now, of course, I've got to check out the PDP-11 simulator. (I'm sure I heard about one that was written in Java, but when I did a search on Google it seemed like every damn CS student in the world has built one as a class project...someone else'll have to provide the link.)

Excellent, ....good code, some of it....

[CDWert]

People are scoffing, but oh what a kludge, caldera only did it for the publicity, who cares.

There is some GOOD code in there, some is crap just like all projects.

There are/were some code segments that were optimized VERY well, and hy dont laugh I remerber USING a pdp-11.

Point is the horsepower ot these machines sucked by modern standards, things we take for granted were MAJOR tasks, some of those routines were refined over a lifecycle longer than BSD and Linux combined.

My dad, a coder starting in the 60 tought me in about 1979, when he handed over his collection of computer mags, Byte and the lot, you can NEVER , EVER have too much source code, good or bad.

Im could care less what Caldera's motives are theyre a dead fish. When was the last time YOU installed that distro ? BUT they should be congratulated, I agree they should open up the other sources as well, but who am I to ask, hey while youre at it all people reading this post should give me all their old ????, see dosent fly. Its theirs , they bought it, and paid for it, sorta.

Im gonna grab it asap, there about 4 things I hate in the 2.4.18/2.5 series kernel, BSD IMHO dosent have the solution in theirs either, I keep looking , Im at a mental roadblock so to say(nothing new there) if these sources even point me in the right direction to solve one of those problems I will be etternally greatful

Re:Excellent, ....good code, some of it....

[Animats]

There is some GOOD code in there, some is crap just like all projects.

Very true. Two real howlers in PDP-11 UNIX:

• Dumb scheduler. Three compute-bound processes near the head of the process table would starve out all other processes.
• Really dumb malloc. O(N^2) speed. It's Knuth's fault. When he discusses memory allocation in the original edition of Fundamental Algorithms, he gives a bad algorithm, says it's bad, and leaves the good one as an exercise for the student. UNIX used the dumb one, right out of the book down to the variable names.

Re:Why is this cool?

[chris_sawtell]

This is a tiny fragment of Multics Source

It's only 20 years old, surely we can find more of it?

Re:Double ewww...SCO/Open Server

[Ethanol]

First off--you're dissing an operating system released by one company in 1993, because a different company has released the source to a different OS, which was written by still a third company, decades before the one you're complaining about. How, exactly, is this even remotely on-topic?

That said, even though it isn't on-topic, I'd like to respond, for personal reasons: I'm one of the half-dozen engineers responsible for maintaining SCO OpenServer. (In point of fact, I'm the one responsible for that DHCP client you mentioned. BTW, if you think it's easy to maintain an entire OS distribution with a team that small, try it sometime.)

I could post voluminous defenses of why OSr5 is the way it is, but won't bother. I'll just say this: Some of your complaints are quite valid. (Others aren't--where in the world did you get the idea we don't have ELF libraries?) But it can essentially all be chalked up to the fact that we're talking about a legacy OS from 1993 which is neither intended to be, nor sold as, a state-of-the-art kernel in 2002. (For that, you want OpenUNIX 8.)

OSr5 is successful in the marketplace because it does what a lot of people need, does it well, does it extremely reliably, and does it in essentially the same way that it's done it for a decade (modulo those changes necessary to ensure that it runs well and takes advantage of the most current hardware)--which means no surprises for resellers and vertical-app vendors. There are fancier kernels nowadays, but nothing else on the market is as stable a platform, and for all its admitted outness-of-date, I'm very proud of it.

To bring this back within hailing distance of the topic: I fervently hope that one day OSr5 will be open source too. I don't really expect it, unfortunately; not all of the code belongs to Caldera. Bits of the XENIX-compatibility code, as you noted, are licensed from Microsoft, and what are they odds they'd ever agree to open-source anything? But it would be very satisfying.

Coolness factor

[dreamsinter]

Hey, for all the "worthless code" grumbles I've read, I think Caldera deserves a lot of credit for this. I think it'll be useful to read and reread this code until I get some idea of how to handle multi-usered, multitasking memory in 64K or less.

That's one thing you won't get from reading code written for systems with >64M.

Re:16bit x86 now?

[LaRueLaDue]

Try finding diskettes for the Microport System V/2.4 port for the 286, or for Xenix for the 286. They actually worked! I would let you use mine, but I inadvertantly tossed them several years ago :-( .

There was something cool about having 5 or 6 people on a 10 or 12 mhz 286 with 6 Mb RAM, actually getting something useful accomplished, and quickly...