I'm pretty sure the jumpers, as you say, just pass through the upper 4 bits of the address bus in the absence of the mapper chip. The CPU has a couple of its External Intructions (CKON and CKOF) decoded to turn the mapper on/off respectively. When the chip is fitted and in the un-mapped mode (normal), it simply passes through these bits. When on, it adds an additional 4 bits to provide a possible 256 4K pages (only 16 can be paged in at any one time) thus providing for up to 1MB of extended memory.
As for the graphics, well I guess you have to be somewhat 'artsy'! The TMS9928/29 VDP was pretty cool, but to do any fast hi-res stuff was quite difficult as the video ram was not accessible directly so you had to transfer it to CPU ram, manipulate it and then send it back via a couple of registers.
The 40x24 text mode was a little limiting, so a bunch of us upgraded to the backwards-compatible MSX Video Chip (the V9938). This was available on a PCB from Maplin for some kit they had at the time. Someone came up with a package that patched Cortex Basic so that you had 80x26 text and some additional high res graphics modes (it had 128K of video ram instead of the original 16K).
I was attracted to this machine for a number of reasons. Firstly, when you powered it up it copies the ROM image into RAM. This makes it so easy to patch/enhance the operating system without having to burn some more Eproms! Secondly, because it was such low volume, there was a small community that, between us, pretty much had to come up with our own hardware and software which forced a learning curve that was really absent on any other platform at the time (and even now). It had a built-in assembler/disassembler and debug monitor where you could set breakpoints, inspect registers and so on. Not to mention it was 16 bits with hardware multiply & divide!
Basically, this machine allowed me to develop skills in assembly language programming and digital design that I have used in my professional life since. I work in flight simulation and, like the aircraft, these machines are around for a long time (a new one today would cost about $12,000,000). Hence it's not unusual to be doing a modification in 32 bit assembler on a super-mini that's circa 1978!
I don't actually know the people involved, I just know their names from looking at the MDEX software sources and documentation. From Googling, John Walker is the co-founder of Autodesk (!!!) and Mike Riddle is also still in the CAD business in Pheonix (bitter split between him and John in the mid-80's). You can hear him talking about the early days using the TMS9900 based system and his SPL language at
http://www.digibarn.com/stories/mike-riddle/MP3_96_mike-riddle1.mp3
There is no way of knowing that Neil Quarmby wrote CDOS. I just happen to remember sending him the cheque for it and getting his TI business card back with the order. He can be found with a quick Google too!
I suppose in light of the fact that I very well could have the only remaining copies of this stuff it would be prudent to back it up to a website/repository somewhere and just make it clear that is why it's there! As long as it's clear that for someone else to use it they should seek the permission of the respective copyright owners, it should be OK? It has no commercial value whatsoever and would be very educational for people to be able to experience what was available back then.
