Talk:StackBeat

t insertion

I'm finding that because of the (pop)>>(pop) semantics, and the fact that the only way to include t is by copying it from the bottom of the stack, I have to use a lot more # than I would if either the semantics were right-to-left or if there was an instruction to push t. Was this intentional? I can see potential to be better for composing than infix notation, if it were but a little bit friendlier. Xifeng (talk) 06:34, 25 August 2013 (UTC)

If you ask, was it intentional to pass the timestamp value only once, then yes. It's implied in the architecture. The processing loop part was designed not to know anything about the stack machine part, and vice versa. It was made in a way that anyone can easily substitute this very stack machine with MSM, STXTRM, Befunge... whatever piece of code that generates a stack-oriented code that takes a timestamp from the stack top and returns a signal level from 0 to 255. Of course, the stack machine part can be extended to initially save the top to a register and then call this register any time. Could you suggest a character name for this instruction? (non-alphanumeric and from the ones not already used) Plugnburn (talk) 16:16, 25 August 2013 (UTC)

That makes quite a lot of sense. I think that _ would be reasonable, since it suggests a fill-in-the-blank. Xifeng (talk) 21:16, 25 August 2013 (UTC)

Updated the article and implementations (with this new _ instruction). I'd also note that in this new, current state, some StackBeat formulas can actually be shorter than their IBNIZ analogs. That's why I'm not going to change argument order anyway. Plugnburn (talk) 17:32, 26 August 2013 (UTC)