Is RPN still relevant?

[Garth Wilson]

(12-18-2023 03:18 PM)bxparks Wrote:  

(12-18-2023 12:08 PM)ijabbott Wrote:  Indeed, a four-level stack is an artifice that has no place in teaching or learning mathematics. The NSTK mode opens other avenues, such as thinking about how computers work internally.

I find the traditional 4-level RPN stack works for me because I can keep track of 4 things in my head pretty easily.

Same here.  I have no need for a four-line display to show what's in the stack registers.  If you can't keep track of them, a display might not be of much use anyway, because it'll show you numbers but not tell you what each number is.

Quote:Beyond 4 levels, [...] it makes programming in a stack-based language like Forth frustrating. (RPL seems very similar to Forth as far as I can tell.)

The following is slightly edited from a stacks article on my website.

When we move to programming, the programmer only has to give his attention to the few stack levels (typically three or less) being used by the particular routine he's working on at the moment; so letting the stack get really deep does not make it unwieldy.  The number of levels that are on the stack waiting for you to get back to is irrelevant, like a ship at sea with no concern for how deep the ocean is at that location as long as it is deep enough to avoid hitting the bottom.  Also, when you call a subroutine which takes, say, two inputs and gives back one output, how many data-stack levels are used inside the subroutines it calls is of no consequence.  It simply doesn't matter (assuming the memory allowed for the stack is enough that you won't ever run out).

Subroutines automatically leave other subroutines' "scratchpad space" on the stack undisturbed.  In fact, in recursion, a routine calls itself, even multiple nested levels deep, which a stack allows it to do.  (As you might expect, it is important to make sure that the condition to back out of the recursion is met before you overrun the available stack space.)

Comments should prevent the possibility of confusion of what's on the stack.  Sure, I've seen very unreadable source code in Forth, but I blame that on the programmer, not the language. I've seen horrid source code in the other languages I've worked with, too.