Stanislav,

This is excellent news. I'm impressed that you are willing to so thoroughly violate this machine in order to publish it's secrets, although, being familiar with your earlier essays, perhaps I should not be so surprised. Honestly, I think I would hesitate. I would like to think I would get down to doing this work once the lispm novelty wore off, but I am not so sure. I wish you the best of luck.

Lacking a physical lispm, I actually started digging into the VLM code with similar intentions before letting myself get sidetracked into hacking Genera. I wanted to learn about the Ivory architecture, produce a better-written software emulator, and (more abitiously for me) reproduce something in an HDL. While there are some world file format differences between the I-machines and VLM, the compiler and resulting machine code are, to my knowledge, the same, and so I reasoned that a working replica could be made given knowledge of the machine code, the emulator's internal CPU representation, and the U-code, the lattermost of which I regrettably do not have. The Ivories weren't nearly as radical a departure from conventional computer architecture as your dataflow CPU would be.

Of course, the result wouldn't really be an Ivory, and would probably be more effort in the end than your direct approach.

My VLM keyboard woes are probably more to do with the silly HHKB I'm using, which feels great to my fingers and works fine for my usual emacs/slime workflow, but whose lack of un-modified F keys is now a serious impediment.

I think I've isolated the problem with Save World, shooting to push a patch and get back to my original task this week... It's quite clear to me that his task would have been much more difficult without an integrated debugger, incremental recompilation, who-calls, etc.