For me it was all about OS-9 at the end

https://en.wikipedia.org/wiki/OS-9

which had a decent C compiler at the very least. I was checking out everything I could about Unix from the public library and creating my own version of as many Unix tools as I could. In terms of OS, the Coco 3 was head and shoulders over anything else you could get up to that point. It even had a windowing system that was a lot like Plan 9.

At some point though I was frustrated with there not being a lot of software for it and I did a data processing job for my Uncle Bob that paid for a new 286 machine although if I knew how much value it made for him I should have asked enough for a 386. The 286 was a massive step up in performance -- in high school I developed a CP/M program for a teacher using a Z80 emulator that was 3x faster than any Z80 could you buy!

> It even had a windowing system that was a lot like Plan 9.

I'd love to see that.

I was a huge fan of the Coco since it was my first computer in 1981 and I progressed from 4K in 1981 to 16K to 64K and from the Coco 1 to Coco 2 with full four slot expansion pack, internal video, modem, keyboard, RAM (256K!) and lowercase video mods and two double-sided floppy drives. I ran a large users group and flew around the country to various Coco shows including the first Rainbowfest. However I have mixed feelings about the Coco 3 because by the time it was finally announced I'd already moved on to the Amiga. I wanted to be blown away and love the Coco 3 but by late 1986 the market was evolving rapidly and the Coco 3 was just "okay". Sure the 6809 was still the best 8-bit CPU ever made and OS-9 was still wildly overpowered but the new graphics and sound were only "okay" and not exciting. If the Coco 3 had launched with those specs a year or 18 months earlier, it would have been compelling.

I knew some of the top developers who had close contact with the execs at Tandy in charge of the Coco product line. They'd been working on an expanded new Coco to ship in 1984 or 85 called the Deluxe Color Computer but it was derailed by two things. First, Tandy kept delaying investing in a new Coco because of the consumer computer crash of '83 and Commodore's cutthroat price-warring. Also, Tandy's much higher revenue PC compatibles were selling extremely well. Pouring scarce design resources into a new sub-$500 non-PC computer just wasn't a priority. Tandy's computer sales people also didn't have any interest in selling low-end models when they were moving dozens of >$1,200 PCs at once to businesses.

The other thing that delayed the Coco 3 was Tandy deciding to rely on Motorola's forthcoming RMS graphics chipset (https://archive.org/details/TNM_Motorola_Raster_Memory_Syste...). The specs of the chipset looked very competitive with up to 32 colors from a 4096 color palette at 320 x 200 and 16 colors at 640 x 400 with up to 8 sprites, horizontal and vertical smooth scrolling and hardware display lists. While demos from the bread board prototype looked good, the chipset delivery dates kept slipping due to bugs in the taped out chip. Ultimately, Motorola ended up cancelling the RMS chipset entirely when they realized that the Amiga was going to be better and the Atari ST was going to be cheaper than 68000-based RMS-based systems.

At that point Tandy had to start over for the Coco 3. They ended up hiring a contractor to quickly design a gate array for graphics and that's how the Coco 3 ended up both late and with less than state-of-the-art graphics. It's unfortunate because an RMS-graphics based Coco with an Hitachi 6309 CPU at 3.58 Mhz would have blown away the Commodore 128 in 1985 and even been quite competitive against the Atari ST in both graphics and CPU speed while selling for less thanks to having a less expensive 8-bit CPU.