The TI-59 had about 1 K of RAM, a magnetic card reader storing 240 bytes per side, and very small ROM modules that could hold about 5 K. TI-59 had all that in roughly the standard calculator form of roughly 6 inches by 3 inches though the battery pack made it a bit thicker than modern calculators. Compare it with an early microcomputer trainer like the Kim-1 and see the difference focusing on designing for use as a calculator versus a computer. An amazing machine for 1977.

Sadly, TI followed the 59 with the TI-88 which was slow, unreliable, and designed to use very bulky add-ons including a cassette interface which was the same size as the calculator. Add the tape recorder and that would be a challenge to use in the field. Most of the TI-59 fan base was hoping for a faster replacement with more memory taking advantage of the improvements to technology.

I have a recent TI graphing calculator and a number of light weight portables but all the extra flexibility comes at the cost of not having quick entry of the fields needed for a calculation.

What made the HP Palmtop computer's calculator so good was the "Solver"

You can customize your HP 95LX by entering equations into the Solver. Then you can solve for any of an equation’s variables using special function keys, one for each variable, that the Solver creates for you. Type in the value, press that variable’s function key, repeat for the other known variables, then press the key for the unknown variable. The Solver does the rest and displays the solution.

The 'special function keys' are just the F-keys already available a the typical top row of keys. Once you have entered and saved you favorite equations in the library, they are avaiable at any time, with the last values used. Makes for quick calculations and 'back calculating'.

I use RPN calculators, mainly HP varieties (they were free at the time, by my brother).

In the end, I wrote my own, by reverse-engineering the HP 15C. It turns out that the core utility of an RPN calculator is something like 10 lines of BASIC. The rest of it is given over to the likes of functions, display, etc.

It worked well enough on a tandy 100. Painfully slow, but it worked. You could change the base in running, quickly display and input decimal x in any mode, and set the size of the circle and logrithm-base (or multiplier) in its memory. Saving eg 126 to the circle register, would make a radian 126 units. Saving -126 would make the circle 126 units.

It's fairly easy to write a 200-digit command-line calculator in REXX, eg the following is a REXX calculator that functions at 200 digits.

You can add your own functions as rexx subroutines, and carry this from OS to OS (rexx is available for DOS, OS/2, Win16, Win32, Linux, and a number of other systems.

At the other end of the spectrum, I do a number of calculations on a slightly modified coin-table, which handles the likes of 1, sqrt(2), sqrt(3) and sqrt(6) exactly.

## Comments

The perfect calculator was the TI-59.

Sadly, TI followed the 59 with the TI-88 which was slow, unreliable, and designed to use very bulky add-ons including a cassette interface which was the same size as the calculator. Add the tape recorder and that would be a challenge to use in the field. Most of the TI-59 fan base was hoping for a faster replacement with more memory taking advantage of the improvements to technology.

I have a recent TI graphing calculator and a number of light weight portables but all the extra flexibility comes at the cost of not having quick entry of the fields needed for a calculation.

You can customize your HP 95LX by entering equations into the Solver. Then you can solve for any of an equation’s variables using special function keys, one for each variable, that the Solver creates for you. Type in the value, press that variable’s function key, repeat for the other known variables, then press the key for the unknown variable. The Solver does the rest and displays the solution.

The 'special function keys' are just the F-keys already available a the typical top row of keys.

Once you have entered and saved you favorite equations in the library, they are avaiable at any time, with the last values used. Makes for quick calculations and 'back calculating'.

In the end, I wrote my own, by reverse-engineering the HP 15C. It turns out that the core utility of an RPN calculator is something like 10 lines of BASIC. The rest of it is given over to the likes of functions, display, etc.

It worked well enough on a tandy 100. Painfully slow, but it worked. You could change the base in running, quickly display and input decimal x in any mode, and set the size of the circle and logrithm-base (or multiplier) in its memory. Saving eg 126 to the circle register, would make a radian 126 units. Saving -126 would make the circle 126 units.

It's fairly easy to write a 200-digit command-line calculator in REXX, eg the following is a REXX calculator that functions at 200 digits.

`/* REXX calculator */ numeric digits 200 parse arge chalk interpret 'cheese =' chalk say cheese exit`

You can add your own functions as rexx subroutines, and carry this from OS to OS (rexx is available for DOS, OS/2, Win16, Win32, Linux, and a number of other systems.

At the other end of the spectrum, I do a number of calculations on a slightly modified coin-table, which handles the likes of 1, sqrt(2), sqrt(3) and sqrt(6) exactly.