RPN83P: RPN calculator for TI-83+ TI-84+ inspired by HP-42S

[bxparks]

(03-14-2024 11:40 AM)Dave Britten Wrote:  Seems like it's working fine on my Pocket. It is, after all, just an 84+ SE in a smaller form factor.

Awesome, thank you for verifying, I'll update my README.md. I mean, sure, that's what Texas Instruments say, but I ain't going to take their word for it without validating. Manufacturers sometimes drastically change the hardware or firmware of a product, without changing the model number or the serial number sequence.

(03-14-2024 11:40 AM)Dave Britten Wrote:  Certainly! On these calculators, you select a category of units (length, area, temperature, etc.), and are given a menu with various units to choose from. To convert, you first press the unit you are converting from, and then press the unit you are converting to. So if I wanted to convert 25 ft to m on the 19BII, I would type 25, press "FT", and then press "M", which would convert the value in the X register. The TI-85 is similar in that you get a menu of units, and press two in sequence to enter a complete conversion function. So to enter the expression 25[ft]>[m], you would type 25, press "FT", then press "M". You can include as many conversions as needed in a single expression.

That's fascinating. I took a quick scan through the TI-85 and 19bii manuals, now that I know what to look for, and I kinda like it. I'm going to think about how this feature can be implemented in RPN83P. No promises though. Quick question: Do these calculators exhibit the roundoff problems in unit conversions shown by Casio calculators?

[Dave Britten]

(03-14-2024 08:28 PM)bxparks Wrote:  That's fascinating. I took a quick scan through the TI-85 and 19bii manuals, now that I know what to look for, and I kinda like it. I'm going to think about how this feature can be implemented in RPN83P. No promises though. Quick question: Do these calculators exhibit the roundoff problems in unit conversions shown by Casio calculators?

They don't appear to. I just tried 1 ft^3->in^3 on one of my TI-85s (ROM 10.0) and I get an even 1728. I don't have my 19BII close at hand at the moment, but I tried on the HP Calc application on my 200LX (basically a 19BII), and also got 1728.


By the way, I don't think I've seen this mentioned yet, but if anybody wants to make RPN83P the primary operating environment on their 83/84, just install TI's Start-Up Customization app:

https://education.ti.com/en/software/det...tomization

It has a few different functions, but one of them is to launch an app immediately at power-on. Just set it to run the RPN83P app, and that's what you'll see whenever you power on. You can still drop back to the standard TI home screen with 2nd Quit like usual, and Start-Up Customization can be easily disabled by running it again if you ever want to revert to the original operation. Great way to get an RPN scientific from ebay for about $25!


Also I was thinking, rather than implement keystroke programming in RPN83P, how difficult would it be to launch a TI-Basic program from within the app? You could simply copy the stack to variables X/Ans, Y, Z, T, and L, and store the registers into a list (e.g. "REGS"), then bring the values back in after the TI-Basic program exits. No clue if the TI OS makes it possible to do this from within an assembly-language app.

[bxparks]

(03-15-2024 12:26 PM)Dave Britten Wrote:  They don't appear to. I just tried 1 ft^3->in^3 on one of my TI-85s (ROM 10.0) and I get an even 1728. I don't have my 19BII close at hand at the moment, but I tried on the HP Calc application on my 200LX (basically a 19BII), and also got 1728.

Interesting. HP and TI must be using higher precision conversion constants than the Casios.

(03-15-2024 12:26 PM)Dave Britten Wrote:  By the way, I don't think I've seen this mentioned yet, but if anybody wants to make RPN83P the primary operating environment on their 83/84, just install TI's Start-Up Customization app:
https://education.ti.com/en/software/det...tomization

Yup, I mention this auto-start capability in my User Guide:
https://github.com/bxparks/rpn83p/blob/d...auto-start

(03-15-2024 12:26 PM)Dave Britten Wrote:  Also I was thinking, rather than implement keystroke programming in RPN83P, how difficult would it be to launch a TI-Basic program from within the app? You could simply copy the stack to variables X/Ans, Y, Z, T, and L, and store the registers into a list (e.g. "REGS"), then bring the values back in after the TI-Basic program exits. No clue if the TI OS makes it possible to do this from within an assembly-language app.

Others have suggested this feature. Various "shell" programs (DoorsCS, MirageOS, Ion) do exactly this, so it's technically possible. The problem is finding someone who has the time, energy, and technical knowledge to implement this. Realistically that "someone" won't be me, at least not within the next ~2 years, but I'm willing to work with someone who has the talent for this kind of thing.

[Dave Britten]

(03-15-2024 03:31 PM)bxparks Wrote:  Others have suggested this feature. Various "shell" programs (DoorsCS, MirageOS, Ion) do exactly this, so it's technically possible. The problem is finding someone who has the time, energy, and technical knowledge to implement this. Realistically that "someone" won't be me, at least not within the next ~2 years, but I'm willing to work with someone who has the talent for this kind of thing.

Understandable, and I definitely wouldn't be able to tackle that either. But it would be a simpler approach to programmability than writing an entire keystroke-programming system, in any case.

[bxparks]

(03-15-2024 03:54 PM)Dave Britten Wrote:  Understandable, and I definitely wouldn't be able to tackle that either. But it would be a simpler approach to programmability than writing an entire keystroke-programming system, in any case.

It actually isn't obvious at this moment that shelling out to a TI-BASIC program is easier. That part of the TI 83 Plus SDK is not well documented at all, at least not to my eyes. Some very smart and dedicated people figured it out 20-25 years ago, but I don't know how much that knowledge has been retained. (Only one of the 3 shells that I referenced earlier makes the source code available.) Interacting with the TI-OS is always a long exercise of trial-and-error, since the official documentation is not always accurate or complete, and the undocumented features are scattered all over the internet.

The advantages of implementing my own keystroke programming is that I know and control the entire stack, up and down. I don't have to work around any undocumented and obscure bugs in the OS. The other advantage of adding my own programming is that there are features that have been added to RPN83P that are not available in TI-OS. The next version of RPN83P will include a bucket load of features which are completely outside of TI-OS.

The best thing would be to implement *both*, in other words, por que no los dos?

[bxparks]

I have released v0.10.0 (2024-03-31) of RPN83P here, https://github.com/bxparks/rpn83p/releases. Download the rpn83p.8xk file, then upload it to your calculator. Here are the highlights:

1) The major feature set of this release are the DATE functions:

* date, time, datetime, timezone, and hardware clock
* add or subtract dates, times, datetimes
* convert datetime to different timezones
* convert between datetime and epochseconds
* support alternate Epoch dates (Unix, NTP, GPS, TIOS, Y2K, custom)
* set and retrieve datetime from the hardware clock (84+/84+SE only)
* display time and date objects in RFC 3339 (ISO 8601) format

The hardware clock exists only on the TI-84+ and 84+SE models, not the 83+ or the 83+SE.

The calculator supports the Proleptic Gregorian Calendar from the year 0001 to 9999, using POSIX seconds (no leap seconds), with timezones as fixed offsets from UTC. The Year 2038 problem is avoided by using 40-bit integer arithmetic for all internal date calculations. If none of that made sense, the USER_GUIDE_DATE.md below has explanations and references for those terms.

See https://github.com/bxparks/rpn83p/blob/d...DE_DATE.md
Apologies for its length, 38 pages as a PDF file with almost 500 screenshots.

2) The behavior of CommaEE button has changed from v0.9, because the comma key is needed to enter DATE objects. The EE function now needs to be entered as 2ND EE, as written on the keyboard labels. There is now a MODE option to invert the behavior of the CommaEE button if you prefer.

See https://github.com/bxparks/rpn83p/blob/d...utton-mode

3) The underlying TI-OS supports 27 single-letter storage variables, named A-Z and Theta. I have exposed these through the STO and RCL commands, using the keystrokes `STO ALPHA A` and `2ND RCL ALPHA A`.

See https://github.com/bxparks/rpn83p/blob/d...-variables

4) I improved the speed of the PRIM (prime factor) function by 40-50% with a relatively small tweak. Here are the running times for factoring the worst case number: 65521*65521 = 4_293_001_441:

* TI-83+ (6 MHz): 20 s
* TI-83+SE (15 MHz): 7.7 s
* TI-84+SE (15 MHz): 9.5 s

See https://github.com/bxparks/rpn83p/blob/d...me-factors

5) The RPN83P app now consumes 3 flash pages (48 kiB). So it takes a little longer to upload to the calculator, especially on the 83+ models.

Hope this all makes sense.

Brian

[jthole]

Thanks! I just upgraded my TI-84+ to 0.10.0.

This calculator is always on my desk nowadays.

The large keys and display make it very easy to use, and your RPN implementation is extremely capable.

Given the low used prices of the TI calculators, this also must be one of the least expensive RPN calculators available

[dmh]

I just installed 0.10.0 and am looking at the new features, looks like you've added a lot of functionality very quickly!

If I key in a date I can save it to a numbered register but not to a named variable. Attempting to save to a named variable gives "Err:DataType". I tried a few different dates including "D{2024,4,1}".

Recalling a date using a named variable works fine. EDIT: actually, this doesn't appear to work or results are inconsistent.

I suspect this is a bug but happy to be corrected if I am not using correctly. EDIT: ok, the documentation only lists that registers can hold date types so not a bug :-)

I'm using a TI-83+ with latest OS.

[johnb]

(04-01-2024 10:04 AM)dmh Wrote:  I'm using a TI-83+ with latest OS.

Thanks for mentioning that as an aside!

It hadn't even occurred to me to check: my Ti-83+/SE is on v1.15 and I see TI has v1.19 available for download.

Finagle's Law* being in full force yesterday, I downloaded v0.9.0 of RPN83P merely an hour or two before Brian posted 0.10.0, so tonight is definitely "updates night."


_________________

*Finagle's Law: "The Universe Tends Towards Maximum Perversity."
Lubarsky's Law of Evolving System Dynamics: "In order to re-can a can of worms, you must use a larger can."
Cole's Law: Slimly sliced cabbage.

[johnb]

Just. Wow.

On the TI download website:

Quote:Why do you need an updated Operating System?

Stupidest question that I've heard all year, but apropos to the date, I suppose...

[bxparks]

(04-01-2024 10:04 AM)dmh Wrote:  looks like you've added a lot of functionality very quickly!

It's funny that you say that, because I feel the opposite. I could be 2-3X more productive in C instead of assembly. For example, it took me 2 weeks (40-50 hours) to translate about 100 lines of C code into Z80 assembly. Granted it was an unusually complex bit of C code, with ~20 equations and ~20 variables, which does not translate into assembly very easily, due to all the hand-written register and stack juggling.

(04-01-2024 10:04 AM)dmh Wrote:  If I key in a date I can save it to a numbered register but not to a named variable. Attempting to save to a named variable gives "Err:DataType".

Sorry, I should have made this more clear: storage variables (A-Z,Theta) are incompatible with Date objects. Only numerical storage registers can be used with Date objects. That's because storage variables can only support native TI-OS objects for compatibility with TI-BASIC, but Date objects are RPN83P extensions which cannot escape the RPN83P environment.

Now, if you have a `D{2024,4,1}` in the X register, and you try to do a `2ND RCL + ALPHA A`, with `A` holding an integer, I don't think I ever tested that... I will have to look at this more closely.

[floppy]

Video (french) https://www.youtube.com/watch?v=juPajoWxEnQ

[bxparks]

I want to mention that there is yet another hardware option for running the RPN83P app, the TI-Nspire with the TI-84 Plus Keypad. I learned recently that the first versions of the Nspire accept interchangeable keypads. If you insert a TI-84 Plus Keypad, the ARM processor starts emulating a Z80 processor running at around 15 MHz, and the calculator essentially becomes a TI-84 Plus Silver Edition with 1.5 MB flash.

Would this run RPN83P? I had to find out. I got one from eBay (for like $19 incl shipping), and yes indeed, RPN83P seems to work perfectly fine! See attached photos.

       

In my personal opinion, this is the ugliest calculator that I have ever owned, the baby blue color is terrible. (There seems to be a rev 2 version of this calculator that is dark-blue or black, with a black TI-84 Plus Keypad. Those are slightly more expensive than the blue version.) But in return for its ugliness, I get an LCD screen that is gloriously HUGE and readable, which is great for people with older eyes like me.

There are multiple versions of the TI-Nspire. RPN83P will NOT work on the ones with a bunch of letters after it: Nspire CAS, CX, CX CAS, CX II, CXII CAS. Those don't accept the TI-84 Plus KeyPad. But the first 2 versions of the Nspire (the blue and the black) are really cheap on eBay. Even with emulation, they are at least 2X faster than the similarly priced 83+ model. The combination of (Nspire + 84+Keypad + RPN83P) may be the best value for a scientific RPN calculator available right now.

[floppy]

(04-02-2024 06:20 PM)bxparks Wrote:  I want to mention that there is yet another hardware option for running the RPN83P app, the TI-Nspire with the TI-84 Plus Keypad. The combination of (Nspire + 84+Keypad + RPN83P) may be the best value for a scientific RPN calculator available right now.

This look better than the TI84-Plus I have. Great Idea.

[Steve Simpkin]

(04-02-2024 06:20 PM)bxparks Wrote:  ...
In my personal opinion, this is the ugliest calculator that I have ever owned, the baby blue color is terrible. (There seems to be a rev 2 version of this calculator that is dark-blue or black, with a black TI-84 Plus Keypad. Those are slightly more expensive than the blue version.) But in return for its ugliness, I get an LCD screen that is gloriously HUGE and readable, which is great for people with older eyes like me.
...

Well, I would say, *almost* the ugliest calculator ever made...
https://en.wikipedia.org/wiki/HP_33s

On a more serious note, I think the work you have done on this project is fantastic!
Thank you

[floppy]

(04-02-2024 06:52 PM)Steve Simpkin Wrote:  Well, I would say, *almost* the ugliest calculator ever made...
https://en.wikipedia.org/wiki/HP_33s

Look at the deep dark suffering in that epic video (HPCC 2020: HP's Worst Calculators (Gene Wright and Włodek Mier-Jędrzejowicz)) https://www.youtube.com/watch?v=BOpzJRDbAnU

[Dave Britten]

(04-02-2024 06:20 PM)bxparks Wrote:  I want to mention that there is yet another hardware option for running the RPN83P app, the TI-Nspire with the TI-84 Plus Keypad. I learned recently that the first versions of the Nspire accept interchangeable keypads. If you insert a TI-84 Plus Keypad, the ARM processor starts emulating a Z80 processor running at around 15 MHz, and the calculator essentially becomes a TI-84 Plus Silver Edition with 1.5 MB flash.

Would this run RPN83P? I had to find out. I got one from eBay (for like $19 incl shipping), and yes indeed, RPN83P seems to work perfectly fine! See attached photos.



In my personal opinion, this is the ugliest calculator that I have ever owned, the baby blue color is terrible. (There seems to be a rev 2 version of this calculator that is dark-blue or black, with a black TI-84 Plus Keypad. Those are slightly more expensive than the blue version.) But in return for its ugliness, I get an LCD screen that is gloriously HUGE and readable, which is great for people with older eyes like me.

There are multiple versions of the TI-Nspire. RPN83P will NOT work on the ones with a bunch of letters after it: Nspire CAS, CX, CX CAS, CX II, CXII CAS. Those don't accept the TI-84 Plus KeyPad. But the first 2 versions of the Nspire (the blue and the black) are really cheap on eBay. Even with emulation, they are at least 2X faster than the similarly priced 83+ model. The combination of (Nspire + 84+Keypad + RPN83P) may be the best value for a scientific RPN calculator available right now.

I really wish the 84 keypad would work with the original CAS model. Strange that TI didn't make them compatible!

I don't mind the baby-blue color, but the way they added alpha keys to the Nspire keypad is really bizarre...

[bxparks]

(04-04-2024 04:08 PM)Dave Britten Wrote:  I really wish the 84 keypad would work with the original CAS model. Strange that TI didn't make them compatible!

I suspect that it was a financial/marketing decision, rather than a technical one. I was away from the calculator world for 25-30 years. I came back to see 25 years of amazing technological improvements... which are thoroughly crippled by the requirements of school exam boards and the narrow curriculum of high school students. It's sad. On the other hand, Texas Instrument is still in the business of making calculators, and HP is not, so I have to give them credit for that.

(04-04-2024 04:08 PM)Dave Britten Wrote:  I don't mind the baby-blue color, but the way they added alpha keys to the Nspire keypad is really bizarre...

There are 3 different Nspire keypads that I see on eBay, the 84+ keypad that I posted earlier, a second one with the alpha keys on the bottom, and third one with the alpha keys off to the side. (See attached photos)

       

Which one do you mean? The one with the alpha keys on the bottom would be 3X more useful in a QWERTY layout, but we can't have that because... School Exam Boards.

[Dave Britten]

(04-04-2024 05:35 PM)bxparks Wrote:  There are 3 different Nspire keypads that I see on eBay, the 84+ keypad that I posted earlier, a second one with the alpha keys on the bottom, and third one with the alpha keys off to the side. (See attached photos)



Which one do you mean? The one with the alpha keys on the bottom would be 3X more useful in a QWERTY layout, but we can't have that because... School Exam Boards.

The one with the tiny alpha keys interleaved with the larger calculator keys is a really odd design.

And yeah, the stupid QWERTY thing is also the reason Casio's fx-CG500 ClassPad is wildly unpleasant for text entry. It's bad enough you have to poke a virtual keyboard with a stylus, but you can't even have it in a normal QWERTY layout.

[floppy]

What are the instructions for installing on "TI-Nspire with TI-84 Plus Keypad (OS v2.46)"
?

So far my findings.

Under linux, a first try with OS 2.54 and tilp was not successfull (the upload of the rpn83p.8xk had no effect: no error, app not seen in the calculator).

Under windows, looks like a student TI-NSPIRE whatever graph&OS SW must be installed, a license key must exists (since I bought one as second hand, no license key so far given) = I am stuck there, too.