Chuck Moore on stack size, stack operators and function arguments

[Garth Wilson]

I enjoy Forth immensely; but although I will be forever grateful to Chuck Moore for it, I have to disagree with him on various things.  Even in this one,

Quote: "A Forth word should not have more than one or two arguments,"

*/ has three, and that's pretty basic stuff.  Two counted strings you want to join may have four stack cells, starting address and count of one, and starting address and count of the next.  I know he didn't like CASE structures; but they sure clarify things if you lay them out correctly.  They may not take more than a stack cell or two when executing, but they can run up quite an inventory during compilation, depending on the number of cases.  I wrote dynamic memory allocator words in Forth seven years ago, ALLOC, FREE, and RESIZE, for non-fragmenting buffers for local variables, including arrays (which ANS's way does not accommodate), and even compile temporary Forth words which could be used to compile other words and then deleted after use without affecting the regular dictionary.  The stack gymnastics to do this were the worst I've ever had (although it took little or no debugging to get it working), and having the local variables would have made it easier.  Chicken and egg!

I don't like his ColorForth ideas either.  I don't like color syntax highlighting except in html where a tag's mate could be at any angle from it, rather than always directly below it or to its right.

I think he recanted his ideas on "The map is not the land."

(I might come back and post again after I've had time to watch the video.)