back
1 comments
This only works well if every verb acts on only one noun. After that, the single-dispatch dot-notation breaks down because you have to pass arguments. There are more sensible ways to write a function call with multiple arguments.

In stack-based languages like Factor, you (very basically) write the arguments in front of the function that operates on them. This works out neatly if things are well-factored. Take the REPL for instance, which in Factor is written as "[ read eval print ] loop", as opposed to "(loop (print (eval (read))))" in Lisp.

But this isn't perfect either. A lot of things just look awkward, possibly more so than with prefix notation. I think there should be some kind of hybrid pre-/in-/postfix notation, where you can write the same function call in multiple ways. You could even split up the function name, for example, "if-then-else(condition, foo, bar)" could be written as "if (condition) then (foo) else (bar)", and under the hood this would be the same thing. I can't think of a better example, but I've done this kind of thing often enough with named arguments.

Of course, all this presupposes that "reading like English" is something to strive for. I think it is, though most efforts have been misguided (Cobol, Visual Basic, replacing punctuation with English keywords).

(By the way, I think the single-dispatch dot-notation that you celebrate as one of OO's foremost innovations to programming, is really broken. I seriously consider it writing things the wrong way :-)