2.11BSD Recovery Project
Almost all of the BSD releases survive to the present. Except one major on: 2.11BSD.
But wait, 2.11BSD is everywhere today. That's true. But it's 2.11BSD pl 470 or some other patched version. The original release has been lost in the sands of time. We have all the patches? what gives?
Why? Well, the 2BSD series was released for the PDP-11. It only ran on the PDP-11 (until modern times when RetroBSD used it to run on MIPS-based PIC microcontrollers). The PDP-11 were low-end mini computers once the VAX were released. They had very little disk space, and tapes were expensive. This meant that tapes were reused, and disk contents were pruned often. Adding to the confusion is that the 2.11BSD tapes were expensive when originally released. One needed to but an AT&T Unix license, which was hundreds of dollars.
So, these factors have conspired to mena that 2.11BSD pl 195 is the earliest surviving 2.11BSD. We have 2.10BSD and 2.10.1BSD, but nothing until the 2.11BSDpl195. A big reason for this is that in the 160s or so in the patch stream the compiler was modernized and nobody wanted the older version because the amount of software that could be compiled was diminishing year by year.
So the goal is to recreate a git repo with all the patches from 2.11BSD as released (or as close as we can get) to 2.11BSD pl 470 (or whatever the latest is). It would be nice to produce boot tapes for 2.11BSD as released, even if they are a bit speculative in places (more on that later)
Why?
But can't you run the patches backwards? We have them all? Why is this project even needed?
Good question. The "patches" in the 2.11BSD series weren't modern unified diffs that retained all the info, even when deleting files. Oftne times, they were shell scripts that removed the files. Or did other things that destroyed information. Destroyed information is impossible to recover, so it's hopeless.
But is it?
Well, to get a 100% perfect recreation is impossible. But how close can we get? That's what this project is doing. We know that 95% of the files are the same as they were in 2.11BSD. We know that many files that were destroyed are the same as they were in 2.10.1BSD, so we can recover them. Others came from 4.3BSD. Only a few files need 'restoration' work.
Status
We've worked our way back to 2.11BSD pl 0. Or at least a possible 2.11BSD pl 0. Now we have to prove it can create a bootable system and all the patches can apply to it to get back to an identical 2.11BSD pl 195 we have. Work is underway to roll things forward in an automated way (starting with 2.11BSD pl 195 and clawing our way back to 2.11BSD pl 0, rebuilding pl 0 and then moving forward).
Wait, why use 2.11BSD pl 195. Can't you use the latest? There's plenty of distributions because of the PiDP-11...
Well. no. 2.11BSD represented a rather eclectic collection of improvements and old-school binary compatibility. A lot of the comatibility was removed from the system (both use as well as support) to slim the system down. As a result, latter-day kernels can't run earlier binaries. So we have to install 2.11BSD pl195. This is a challenge because the 2.11 instructions for doing this are hard to come by (but at least possible). And this pre-dates disk labels, so you have to learn about old-school partitioning schemes.
So why not 2.10.1BSD? That might be possible, but is tricky: the system call interface changed radically betweent 2.10.1BSD and 2.11BSD, so initital attempts have hit roadblocks since a number of tool-chain components also changed formats.
So there' sa lot ot do still to prove this out, and I need to write the paper to go along with it.