> As an HP calculator lover (I wore the ENTER > = t-shirt 40 years ago) I understand the stack and RPN but have no idea how I'd write a trig or stat function if it weren't already built in!
From talking to several people who've implemented things like this, I think they would've said the same thing before.
Nobody walks around knowing how to implement these kinds of things. If you need it, you just start trying, learning, reading, etc, until eventually you do it.
Anecdote about people at different stages of that learning:
I worked at FORTH, Inc. one summer in the 80s before starting college. They sold a Forth system which they were converting to a new language standard, which (among other things) had changed the canonical booleans from 1-for-true to -1. One day I noticed the trig functions were giving wrong answers.
So I bring up the code with my boss, and it's pretty hard for us to figure out -- compact, no explanation, tables of random-looking numbers. Someone stops by (Dean Sanderson? iirc), we tell him what's up, he looks at the screen, and almost right away he's like "Oh, this is easy! All you need to do is swap the lines in the tables."
FWIW, I did try. Many languages now have websites where one can experiment with an interpreter or compiler. I've used them for playing with JavaScript, APL, Python, etc. Strangely, there does not seem to be one for FORTH. SwiftForth from forth.com does not work with 64-bit Intel Mac OS-es like Catalina which is what I have [1].
I tried gForth as well but during the installation, it invoked HomeBrew which gave a series of hostile messages that my system is obsolete and to not bother asking for any help or support. See [2] and the output provided by the poster - it is what I saw in my case.
From talking to several people who've implemented things like this, I think they would've said the same thing before.
Nobody walks around knowing how to implement these kinds of things. If you need it, you just start trying, learning, reading, etc, until eventually you do it.