Crosscut

At this point, I'm convinced that postfix syntax provides little to no advantage without stack-based evaluation. But to drive that point home, let's look at the last in my list of reasons for switching to prefix syntax: The "series of statements" approach brought about by using algebraic effects.

Because if you're going to use values to track side effects, in the form of monads or maybe using linear types (as I had previously planned), then every function body ends up as one big expression. And then backwards prefix code that does not match the order of execution, is a possibility everywhere. If postfix syntax has an advantage, it's in that scenario.

I think I've already shown that it probably doesn't. But even if I'm wrong, if I'm missing something, using algebraic effects changes the equation. Because we do end up with a series of statements, and those statements are in the order in which they get executed in. Which further erodes any advantage that postfix syntax could have.