Art,
Yes I was unaware that all the operations you cite was still in Mach 3.  I used to build a low class robot for the commercial greenhouse industry.  It used a 6809 microprocessor at 2 Mhz, 32K eprom, and 64K ram. All the programming was done in Assembler and burned to the eprom. It also had a speech synthesizer and PWM for the motor controller so there was a lot going on. When my programmer got hired away after 4 years we turned the programming job over to the company making the boards.  They were shocked when they printed out the Assembler code and it was over 40,000 lines! Fortunately he had done a great job thoroughly documenting nearly every line of code and they had no problem working with it. Production on that control system went on for about 15 years and ended when some of the critical chips were no longer available even on the surplus market. At the assembler level you can do an amazing amount of work with very few resources. Extremely time consuming though.

For a plasma cutting machine, you need forward and reverse on contour, you need jog off contour and either relocate or return to contour, and you need a kerf compensation routine that works, mach3 doesn't.

Fascinating thread, even if old, but I will vehemently DISagree with some of the suggestions.
NO #var
No SUbs
No parametric programming
Any of these would make it NO DEAL for me.

No wizards
Why??? They can be useful. Imagine having to replace the Text Writer!

Conditionals would be nice, but not if they brought in a lot of bugs.
G65 ... a bit of a can of worms I think, although the basic idea is interesting. But the idea that G65 would stomp on some #var variables - VERY debatable.  OK, you would just have to skip using the first 30 #vars in any program.

The whole idea that any small changes in dimensions or cutter sizes should be dealt with by going back through the entire CAD/CAM/Postprocess/check cycle is ridiculous - especially as you can deal with any of these changes by changing a single #var parameter (or even by cutter compensation). Current CAD/CAM is just too mindless to be useful for many of us. Until you get to the $50k - $100k level.

The idea that hobbyists don't need anything more than Start/Run/Stop is amusing but also commercially stupid and very insulting. If someone becomes serious about their hobby, which is normal, they are going to want to do a LOT more than that.

Rhubarb rhubarb rhubarb.

Cheers
Roger