The Museum of HP Calculators

HP Forum Archive 08

[ Return to Index | Top of Index ]

RPN as a language?
Message #1 Posted by David H Thompson on 22 Aug 2002, 6:14 p.m.

I was surprised to read Michael Meyer's statement that RPN is a language. Surely it's a notation. Even if we rise to the next level of HP complexity, RPL, that's still not a language.

Please correct me, but I thought that most people had now come to accept that computer "languages" are nothing of the sort: they are all just notations. I thought that Knuth and others had long since settled this matter.

Re: RPN as a language?
Message #2 Posted by Michael Meyer on 22 Aug 2002, 6:52 p.m.,
in response to message #1 by David H Thompson


You are, of course, correct in the literal sense, and I appreciate your pointing this out. Technically, there is no official "language" of law, chemistry, etc. My point was that, in my opinion, the way we frame things can affect our abilities to solve problems. In this case, I believe RPN "teaches" the mind to work problems logically and better than AOS. With AOS, just entering the problem doesn't reinforce the mechanics of solving the problem. And that's only my opinion. I also believe that schooling, in general, teaches as much in the way of how to think as it does a mass of specific facts. For example (in my opinion), learning a foreign language does more for a person than just giving them another set of words to use. I believe that it can teach the person a process of learning that can extend to other areas of thought. I also believe it gives a structure-- both good and bad-- with which to enhance thinking, as most of our thinking is done in a "language" (certainly the verbal thoughts). I once heard a quote that makes sense to me, that, "we build our environment and it then builds us." I believe there is truth to this.

Anyway, the main point is that I believe the learning and use of RPN gives a person more than just a faster way to solve problems. So, this isn't a correction.... I believe you are correct. I was using "language" in a more general term for a process (or notation) that also influences the logic and thinking of the individual. I apologize for my general and perhaps sloppy use of the term....

But I can't help but ask: how would you explain the term, "the language of love"? <grin>


Re: RPN as a language?
Message #3 Posted by Ellis Easley on 23 Aug 2002, 6:39 a.m.,
in response to message #2 by Michael Meyer

Michael wrote: "With AOS, just entering the problem doesn't reinforce the mechanics of solving the problem."

I agree, AOS with its emphasis on entering the equation the same way it is written seems to be organized around the idea of entering problems from a textbook. The machine, and whoever wrote the lesson, will solve the problem - the user is just there to press the keys. I guess once you get to the point that you can write your own textbook, you can solve your own problems!

I recently got a TI SR-50A with its manual. This calculator is the second version of TI's equivalent to the HP35. Near the end of the manual there is a four page appendix entitled "Register Level Processing". It explains that internally, the machine has three registers - X, Y and Z. Their use is pre-programmed in the following way: X and Z are used for "+" and "-", X and Y are used for "*" and "/" and exponentiation, and X alone is used for single argument functions. When you hit a function key, the contents of X either stays put, is moved to Y, or is moved to Z, depending on the function. This model has no parentheses and apparently they split (+,-) and (*,/) so they could do "sum of products" without manually saving the intermediate results. I wonder if TI provided this explanation just to parallel the one in an HP manual, where it is necessary to understand it to use the machine. But as an RPN user, it is frustrating to know that there is a three register structure in there that I can't use as a stack!

Re: RPN as a language?
Message #4 Posted by Chan Tran on 23 Aug 2002, 10:01 a.m.,
in response to message #3 by Ellis Easley

I will have to look at my TI SR50 again. I thought it has parentheses. Without the parentheses it would be a lot less than the 35 (assumed that the user prefers AOS)

Re: RPN as a language?
Message #5 Posted by Ellis Easley on 24 Aug 2002, 8:58 p.m.,
in response to message #4 by Chan Tran

I have the SR-50, SR-50A and SR-51A but I only have the manual for the -50A. None of them have parentheses. The SR-52 and -56 do have them.

The "sum of products" system must have been TI's first attempt to get around RPN. It reminds me of the "and-or-invert" arrangement used in programmable logic devices. Some of the examples in the book require re-arranging things. In fact, here's what the first page says:

"The SR-50A uses the algebraic entry method to simplify data entry into the calculator. For simple problems, the numbers and algebraic functions are entered into the calculator in the same sequence as they are stated algebraically."

Note the qualification - "for simple problems"

The SR-50 and -50A have the same key layout but the difference is more than just the style of the case. They have different PCBs. However, they both give the same result for the "calculator forensics (I think that's what he called it)" benchmark test posted here by Mike Sebastion (I think that was his name): (in degrees mode) 9, sin, cos, tan, atan, acos, asin. The result is ideally 9. I go a step or two further and subtract 9 from the result and multiply by 1,000,000. This lets me see the guard digits (if there are any) and also gives a sort of "parts per million" result.

BTW, the SR-50, -50A, and -51A have a very good score on this test, 4.661314 "ppm". Only some of the later HP's give a better result among my calculators - HP 71B, 48SX, 48GX, 32SII and 20S all give -1.35733. HP 67, 34C, 41C and 15C all give 417.403. HP35 (original ROM) gives 2983.113, HP45 gives 4076.644, HP25 gives 4076.649. TI SR-52 and -56 give nearly the same number as -50 except there appears to be one less guard digit. The main purpose of the test, according to the guy, is to identify calculators that may have the same firmware inside, but it also gives some indication of the precision an/or accuracy of a calculator.

Re: RPN as a language?
Message #6 Posted by db(martinez,california) on 24 Aug 2002, 11:54 p.m.,
in response to message #5 by Ellis Easley

the "winner?" and new champ for ellis's and mikes calculator forensics score is the novus mathematician with a result of 156238. that may sound bad but it's probably one whole order of magnetude better than you would get on a slide rule.

Re: RPN as a language?
Message #7 Posted by Ellis Easley on 26 Aug 2002, 12:31 a.m.,
in response to message #6 by db(martinez,california)

Some of my TI's are a little higher - SR-40/TI30 (same thing) give 177087.1, "Slimline" TI50 (an early LCD) gives 177087.09 - so choose the "Slimline"! Other TI numbers: TI55 7726.486, TI57 4746.3834, TI25X solar 52.205, TI34II 3.512065, TI66 (LCD version of 58/59, built by Toshiba) 2.294775, TI34 -1.36296 - exact same result as HP6S. This shows the value of this method for identifying calculators that are the same internally. That these models are the same was pointed out here, I think it was during a discussion about calculators used in algebra classes and how to interpret sqrt(-9) (or was it -sqrt(9)?) When you have them side by side knowing they are the same functionally, it is striking how much better TI arranged the keys than whoever designed the HP6S - although the 6S is faster, very obvious on the trig functions (if that matters on a non-programmable). TI replaced the 34 and another model with the 34II, with two line display and menus - it takes multiple keystrokes to get to the trig functions on this one.

Re: RPN as a language?
Message #8 Posted by Marc Brucker on 23 Aug 2002, 2:37 p.m.,
in response to message #2 by Michael Meyer

I agree with Michael. I have been lurking here for some time, and I have learned that RPN is a useful tool for "getting my mind right", in addition to solving the problem at hand. I am but a noviciate to HP calcs and RPN, but I can already feel my Really Powerful Neurons doing much more work than with the algebraic system of entry logic. By the way, I found a calculator program (assembly, I think) online at a TI calc site that modifies my TI-86 to RPN. Does anyone have experience with these types of programs? I will post the link if anyone is interested. Thanks to all!

Re: RPN as a language?
Message #9 Posted by db(martinez,california) on 24 Aug 2002, 7:35 p.m.,
in response to message #8 by Marc Brucker

yes, please do post the link. someone else mentioned a program like this about a year ago. what a great (and subversive) idea.

Re: RPN as a language?
Message #10 Posted by Marc Brucker on 25 Aug 2002, 10:38 a.m.,
in response to message #9 by db(martinez,california)

Here goes the link to the subversive TI-86 RPN utility. I believe that when downloaded as a ZIP file and extracted, the source code for the program is displayed as another ZIP file to be extracted. Perhaps the program can be ported to other calcs. Go here:

I have not put it on my calculator yet, but plan on doing so, soon.

Re: RPN as a language?
Message #11 Posted by Vieira, Luiz C. (Brazil) on 24 Aug 2002, 8:45 p.m.,
in response to message #2 by Michael Meyer


I wrote this and, even knowing that the subject in here is not the same, the contents are related with each other.

Just my US$0.05...

[ Return to Index | Top of Index ]

Go back to the main exhibit hall