I remember the Unix wars as more ...
You are right. I think I described the first battle of the Unix wars above. You are describing the 10th or 100th battle. Through all of it ran a few undercurrents: Berkeley versus AT&T, workstation versus server, academic versus full commercial, the battles between various CPU families (of which there were many more in those days, Prime and NS 32x32 and Data General and ....).
If you want to see the Unix war between AT&T and BSD in a nutshell, look at the transition Sun did: SunOS was based on BSD, and then in the upgrade to the infamous SunOS version 5, users suddenly figured out that they had suddenly transitioned to SVR4, and everything stopped working. It was either a great victory or a disaster, depending on your viewpoint.
Funny anecdote: A friend worked at a research lab which had only a single VAX. They wanted to run both VMS and Berkeley Unix on it. But it had only one disk drive. So every day at noon the disk pack was replaced, the computer rebooted, and it has one OS in the morning, another in the afternoon. Since there were no other storage media, it was very hard to transfer data between the OSes. Another interesting example: I was working with a big commercial enterprise which was "true blue", using only IBMs for their data processing at a giant production plant and warehouse. But they had one of the first fully automated robotic warehouses, and the robots were controlled by a few PDP 11-70s. Initially, they ran RSX-11 on them, but after being unhappy with it, replaced it with Unix (and I don't know whether they got a research version from Bell labs or from Berkeley). Imagine doing industrial control in Unix in the early 80s! Then they had to upgrade it to a VAX, and didn't know whether to move to VMS or stay on Unix.
In the early 90s, I was working (not as a computer scientist, but still doing research), using big mainframes and VAXes for data processing, but also lots of Unix machines. It was a mix of IBM, HP, NeXT, Sun, and one SGI in my wife's office. Compatibility between OSes at the source code level (both in shells scripts and in C code) was AWFUL, and everything was full of #if statements. Makefiles look like they had come from the salad bar in the cafeteria. That's when I learned to be super systematic about isolating as much of machine dependencies in a single place as possible. And then, about 96 or 98 POSIX started being a real thing, with OSes actually being compliant enough. I stopped using OS-specific documentation, bought two POSIX books (one for 1003.1 and Bill Gallmeister's POSIX.4 book), and used them for programming references. This made life somewhat easier. A few years later I actually met Bill, he was also involved in high school band as a parent (our kids are about the same age). Very nice guy.