History of this Machine
Running IXI's X.desktop Software

I got this machine in the early half of the 1990's whilst working at a computer surplus / resale / repair place in Cambridge. About 40% of the equipment that came in *wasn't* a PC back then of course, making things a lot more interesting.

I was collecting a few old 8-bit micros at the time, but really wanted to get hold of a 'proper' computer - a Unix workstation of some sort. Chances of that happening were around zero I thought; anything capable of running Unix generally had a lot of memory and disk and so commanded high prices.

Then this Tek XD88 showed up. It was lacking a keyboard, the disk was throwing up hardware errors, and it came with no documentation or software other than what was on the failing disk. Covers off. I've removed the framebuffer board in this photo; normally it sits upside-down between the CPU board and the top of the case.

Backing up original (failing) disk image

Luckily the owner of the surplus place said that I could have the system itself - a dead disk was no use to him, and nor was 16MB of proprietary Tek memory that wouldn't fit anything else on the planet. I had to leave the nice 19" Tek screen as he wanted to sell that for serious money though.

I gradually collected bits over the next couple of years or so - someone found a keyboard being thrown out at Cambridge University for me (implying they must have had an XD88 at some point too), a nice 16" Trinitron screen came from my own university, and a guy in the USA kindly sent me a mouse.

Loan machine?

Turns out the machine had belonged to IXI and looked to be a development box for their X.desktop software; I coaxed the disk into enough life to boot - albeit with lots of problems due to missing files in dead areas of the disk. The 'pool' sticker above implies that maybe the machine was on loan from Tek; perhaps they'd allowed IXI to use it in return for having this hot new desktop software ported to their platform?

I managed to copy the raw image of the failing hard disk to a more modern 1GB drive, then boot from that and fix all the filesystem problems; the Tek didn't care about being on a different geometry drive.

From there I found a description of the partition table and boot blocks within one of the system C header files - which meant I could build my own to take advantage of the whole of a 1GB drive and create filesystems on it. I copied the system's diagnostic / boot partition (which is about 3MB) across from the 300MB drive image to a 1GB drive such that it was exactly the same place and so the system would find it on startup. Then I could use fsdump / fsrestore to copy all data across from the 300MB drive image to the new and larger filesystem on the 'new' 1GB drive. Amazingly it all worked, leaving me with a working system that was using the whole of a newer disk.

Then it was a case of building a few useful demo bits and pieces - bison, gcc, xearth, xfractint, povray etc. (I can't get bash to work yet, probably because I'm using a really old version of bison)

[next: Useful info]