What's cool about this System RPL program that solves the quadratic equation? It's a slighly modified version of the program on page 53 of An Introduction to HP 48 System RPL and Assembly Language Programming by Jim Donnelley (page 63 of the PDF version on hpcalc.org). The change is that it includes the inputs (%1 % -4 %3) and a non-standard "showStack" word at the end:

DEFINE a 6GETLAM

DEFINE b 5GETLAM

DEFINE c 4GETLAM

DEFINE root1 3GETLAM

DEFINE root1STO 3PUTLAM

DEFINE root2 2GETLAM

DEFINE root2STO 2PUTLAM

DEFINE Subr 1GETLAM

%1 % -4 %3

::

    0LASTOWDOB! CK3NOLASTWD

    CK&DISPATCH0 3REAL

    ::

        %0 %0

        ' :: a %2 %* %/ ;

        {

          NULLLAM

          NULLLAM

          NULLLAM

          NULLLAM

          NULLLAM

          NULLLAM

        }

        BIND

        b DUP %* a c %* %4 %* %-

        DUP %0< casedrop

            :: "Complex Roots" ABND ;

        %SQRT

        b %CHS OVER %+ Subr EVAL

        root1STO

        b %CHS SWAP %- Subr EVAL

        root2STO

        root1

        root2

        ABND

    ;

;

showStack

HALT

The answer is that it runs on my 32-bit RPL system written in C++.

So far the system supports BINTs, secondaries, LISTS, REALs, STRINGs, loop environments, temporary environments and local variables (LAMs). It does error jumps, and argument validation. The garbage collector can run incrementally.

I've taken a few shortcuts of course, but only a few. REALs are currently binary instead of BCD. This let me get them working quickly and it can be fixed later without affecting the rest of the system.

I'm still trying to figure out a way to deal with low memory conditions. Real RPL does garbage collection as soon as it needs to, which means that objects can move around almost any time. The garbage collector fixes RPL pointers but it can't fix C++ pointers that may exist. Dealing with this in C++ would be very error prone.

I should be writing Prime programs but this is just way too much fun :)

Dave

Cool!

>but it can't fix C++ pointers that may exist.

Good luck... :-| (this is where it all falls apart)

TW

Edited: 10 Nov 2013, 12:11 p.m.

Quote:

---

Good luck... :-| (this is where it all falls apart)

---

I have a couple of ideas that might work. I'll let you know how it goes.

Dave

16 Bit Windows uses handles instead of pointers. All APIs use these handles which allows for moving around the associated memory objects.

If you store a reference count in such a handle, removing unreferenced objects becomes a little easier. Objects containing handles to other objects need a recursive approach for the garbage collector. This is what Java seems to have implemented.

Thanks for the tip Marcus. I'd forgotten about that approach.

Dave