Bootstrapping landscape alphanumeric RPN

[Joseph_21sv]

(01-06-2015 07:34 AM)Didier Lachieze Wrote:  I've voted for A as I think it's the simplest and fastest way to bring your ideas to life, but I have already on my Android phone a working landscape alphanumeric RPN calculator, and a very good one: Free42 !



I would be interested to know what you plan to do differently/better.

Free42 is an obvious candidate for porting to real hardware, all you need is to find someone to 3D print the calculator body and ROM to burn it into and RAM to store user data and a microcontroller to run the machine. With the deflation of computer memory cost since 1995, a calculator which would sell for 120 1988 dollars, would have almost four times the memory capacity of the original HP 42S, if no more. Assuming the quadruple RAM hack is still going to be possible, you get ca. 32 1/3% of the proposed memory space of the WP 43S potentially available. Speaking of the WP 43S, isn’t its target production price slated to be close to 120 1988 dollars looking from the perspective of total market value of the proposed production run anyway? Not to detract from Free42, but to sell it in a physical housing at an essentially equal price to a machine so vastly superior to itself as the proposed WP 43S—never mind that the two will not be in the same orientation—should not but inadvertently kill this plan. And inadvertently killing it means we start back at square one with the bootstrapping stage of this project.

(01-06-2015 08:57 PM)sa-penguin Wrote:  Option A is a good start.
Using something like PhoneGap, you can create HTML-based screen layouts. For those demanding a particular keypad layout, this allows you to play with positions and customise.
It also allows you to do things you might not want in the real model, ie. custom keys (do you really want a touch screen on a physical calculator?).

Other benefits involve things that would be mighty hard to implement on a physical machine. Imagine pressing the function key - and suddenly, ALL main text on keys gets "swapped" with the secondary function text. [The only way I can think of doing this on a physical board requires backlighting LEDs, and sheets of Perspex acting as "optic pipes"]

Somebody did really want a touch screen on a real calculator four times: the little known Sharp ÉL-9600 series, the Casio ClassPad series and the HP Prime and never-produced HP Xpander have one.

Have you read of LCD and OLED keyboards? [the only way you can think of—I almost hate to say it—looks like making life hard for yourself by comparison to these]

(01-07-2015 01:05 AM)Jim Horn Wrote:  Oh, good heavens, don't use the HP-67 AOS program! It was done to prove it was possible but it's impossibly slow. I just wanted to see that a full ~13 level parenthesis implementation could be done on the '67. With its limited speed and memory, the results work but could take many seconds per key press.

I do second the suggestion to work with Free42 or the WP-34S simulators. That way you can concentrate on key labeling and use rather than the heavier duty work of actual functions, most user interaction, etc.

Best to you!

Of course not on the original ’67, especially now that that would be ridiculously anachronistic anyway. However, a landscape hardware emulator of it with a modern CPU and an alphanumeric red OLED display would run that program so much faster and also be able to display the program steps as mnemonics and display the algebraic expressions as the user enters them in.