Branching would be something like:

  case JMP: {
    ip = program[ip + 1] - 1;
    break;
  }

Hi. I updated the article to include a bit more information on registers/branches, and an example on the github with a loop. I wanted to leave the registers (or well the instructions for moving/setting registers) as an exercise for the reader to see if they could implement it themselves :)

You could implement looping by implementing an instruction, say 'JMP', that takes an 'address' as a parameter that modifies the ip. Branching can be implemented similarly with an instruction called 'BNZ', which takes two parameters, a register and an address, which changes the ip to the given address if the register is non-zero. SUB would be useful so you can subtract a register from itself to branch if equal. Other routes include using the top of the stack with your operations and branching, so you can stick to one parameter in the above instructions.

Agreed. I wrote a simple virtual machine, and the most interesting thing was writing the compiler that would read my "assembly", and spit out the bytecode. Including handling labels and jumping instructions.

My VM was very simple, and I've not touched it for a while, but the whole thought of how to handle branching was what made it more interesting than other similar toy systems.

https://github.com/skx/simple.vm

This _is_ a toy system, but "No looping or branching" does not imply "toy system". Counterexample: http://dtrace.org/guide/chapter.html#chp-intro-6