by Carson Reynolds
Made it to section 1.4 (Computing Actions) in The Structure and Interpretation of Classical Mechanics which I think I’ll call SICM (read sick’em boy) for brevity. Figuring I needed a refresher on scheme and the bizarre editor they packaged with it, I spent some time trudging through the Appendix on the language.
In the midst of it I started thinking about massively parallel virtual machines. Strange ideas like having one virtual machine for each closure. Or maybe one real machine for each closure. Maybe something like this: each procedure gets its own reconfigured architecture that is optimized for that particular operation. Compilation essentially becomes creating hardware specifications for a particular function of code. There may be some interesting analogs with things like the raw architecture, where each little processor gets a ( ).
I installed their mechanics environment on my laptop. I’ll have to postpone grand ideas about reconfigurable systems in a turtles-all-the-way-down universe. In the mean time, I’d like to wade through the juicy bits of SICM before monday.