NumWorks open source calculator

[Chasfield]

The calculator forensics calculation returns 8.966758 (for 9 degrees) on this device and it seems to always display results to 7 significant figures. Calculation precision seems therefore to be on the low side.

Is it perhaps squandering higher levels of precision by carrying forward rounded up/down display values rather than underlying register values?

[Namir]

(09-06-2017 05:59 PM)Chasfield Wrote:  The calculator forensics calculation returns 8.966758 (for 9 degrees) on this device and it seems to always display results to 7 significant figures. Calculation precision seems therefore to be on the low side.

Is it perhaps squandering higher levels of precision by carrying forward rounded up/down display values rather than underlying register values?

I tested the expression:

(355/113 - pi)/pi * 1e9

I got a unit 2 days ago!! I was able to update the firmware version this morning.

The new firmware version of 1.1.0 gave 84.91xxxx which is good. The previous firmware version of 1.0.3 gave a higher number (did not think to record it) which reflected poorer accuracy. SO it seems the new firmware version has enhanced accuracy.

Namir

[Namir]

Where is the detailed manual for using the NumWorks calculator? I looked online and the User manual was a single web page. Did I miss a download link somewhere?

Nam9r

[EugeneNine]

now grab the x48 code from sourceforge and port it over

[Luigi Vampa]

(09-06-2017 08:08 PM)Namir Wrote:  (355/113 - pi)/pi * 1e9

I got a unit 2 days ago!! I was able to update the firmware version this morning.

The new firmware version of 1.1.0 gave 84.91xxxx which is good. The previous firmware version of 1.0.3 gave a higher number (did not think to record it) which reflected poorer accuracy. SO it seems the new firmware version has enhanced accuracy.

Namir

... some more digits from Free42 ;0)

[Joe Horn]

(09-06-2017 10:21 PM)Luigi Vampa Wrote:  ... some more digits from Free42 ;0)

Not bad, but the final 8 digits are incorrect. HP 50g with LongFloat installed yields as many digits as you ask for:

84.913678658370862584021992894198817243148...

[ijabbott]

(09-06-2017 05:59 PM)Chasfield Wrote:  The calculator forensics calculation returns 8.966758 (for 9 degrees) on this device and it seems to always display results to 7 significant figures. Calculation precision seems therefore to be on the low side.

Is it perhaps squandering higher levels of precision by carrying forward rounded up/down display values rather than underlying register values?

One bug is that the 'ans' memory re-parses the displayed textual output of the last answer as its input. So for example:

1/9*9 = 1
1/9 = 0.111111
ans*9 = 0.999999

[Hubert Weikert]

(09-06-2017 08:10 PM)Namir Wrote:  Where is the detailed manual for using the NumWorks calculator? I looked online and the User manual was a single web page. Did I miss a download link somewhere?

Nam9r

They have it currently fully available only in French, please see
https://www.numworks.com/fr/ressources/manuel/

Hubert

[Eddie W. Shore]

(09-06-2017 08:10 PM)Namir Wrote:  Where is the detailed manual for using the NumWorks calculator? I looked online and the User manual was a single web page. Did I miss a download link somewhere?

Nam9r

I didn't see a full manual on their web site either, which is troubling.

[Hubert Weikert]

Got my device yesterday, and looked 'under the hood'...

(08-29-2017 03:57 PM)mfleming Wrote:  The last page of the schematics indicate connectors for QUADSPI, USART, SPI and microSD. No mention is made of any of these connectors on the web site, so I'm sure they're not populated . . . but could be?

What is called USART (U7) is a 64 Mbit flash memory, wired for QuadSPI. The pads for are on the board, same as for the microSD (U8) with resistor/FET (R15/Q4), all of them are not populated.

(08-29-2017 03:57 PM)mfleming Wrote:  The processor supports USB OTG, but the USB ID pin is left floating (pity!) so it can only be used in device mode. I wonder if the pin is brought out to an appropriate TP?

All five USB pins are available directly at the USB connector. So it may be possible to connect the respective microcontroller pin to the USB connector with a very thin wire.

I made some pictures of the board, they are available at Pictures Calculator

Hubert

[Anderson Costa]

Tested some number expressions on NumWorks. The software is very limited to execute math operations. Tested ln(-1) and gave undef as a result instead of i*pi. Is any English manual available for this calculator?

[toml_12953]

(09-11-2017 01:06 AM)Anderson Costa Wrote:  Tested some number expressions on NumWorks. The software is very limited to execute math operations. Tested ln(-1) and gave undef as a result instead of i*pi. Is any English manual available for this calculator?

The only manual I could find is a one-page, useless document at

NumWorks Manual

I was thinking of getting a unit but the software is very immature and the Python implementation is a joke. I'll keep an eye on it, though, since it may get better in the future if the creator can keep from going bust.

[pier4r]

I am not sure whether numworks uses micropython or only the python syntax (as the Cas engine on the prime).

Anyway as far as I know mycropython is available via ndless (so, unofficial) on the nspire series and it will be available also on at least the fx-CG50 of the casio series.

https://micropython.org/

http://community.casiocalc.org/topic/763...entry61644

https://tiplanet.org/forum/viewtopic.php...23#p228623

And why do I report this? Because if the various implementation of micropython are somehow compatible with each other, micropython has no extensive math library optimized for low ram environments (numerical library, CAS is nice, but not stricly necessary. We have still our brain). Anyway maybe the casio - ti - numworks community may slowly build one. Although if the effort is not orchestrated, it will take pretty long time.

[debrouxl]

(03-20-2018 10:23 PM)pier4r Wrote:  I am not sure whether numworks uses micropython or only the python syntax (as the Cas engine on the prime).

Numworks uses MicroPython, as can be checked in the source code of the calculator's OS: https://github.com/numworks/epsilon

Bernard Parisse's work for a Python-like syntax in giac was a great idea, quickly executed, and HP was fully right to integrate it into the Prime's OS... but I have a hard time considering it as more than a stopgap measure, especially now that Casio decided to integrate MicroPython to the fx-CG50 / Graph 90+E.
Although they have failed to do so yet, despite years of hints directly delivered to management, TI will eventually have to settle down on some Python implementation as well (well, I really hope so, but the reputation damage caused by not including Python is too high anyway, and will end up reflecting badly on sales at some point)... and as far as the 83PCE & 84+CE's weak hardware characteristics are concerned, MicroPython would be the only embedded implementation which makes sense. In 2010-2011, I worked a bit on porting
Therefore, if they haven't started doing it yet, HP should really think about integrating MicroPython, and an associated development environment (they already have a program editor and a debugger !), into the Prime, by the end of 2018, so that it's available for the 2019 exam seasons (in both hemispheres).

[toml_12953]

(03-20-2018 11:08 PM)debrouxl Wrote:  [quote='pier4r' pid='93438' dateline='1521584632']

Bernard Parisse's work for a Python-like syntax in giac was a great idea, quickly executed, and HP was fully right to integrate it into the Prime's OS... but I have a hard time considering it as more than a stopgap measure, especially now that Casio decided to integrate MicroPython to the fx-CG50 / Graph 90+E.

I've looked for more information on Casio Python but haven't been able to find any. Do you have a URL?

[debrouxl]

(03-21-2018 08:43 AM)toml_12953 Wrote:  I've looked for more information on Casio Python but haven't been able to find any. Do you have a URL?

I haven't looked for resources written in English, but in French, there are TI-Planet topics, the first of which was linked by pier4r above, where critor details information found in, AFAICT, public information sources: https://ti-pla.net/t21189 , https://ti-pla.net/t21193 , https://ti-pla.net/t21198 (I used our URL shortener).

Bernard Parisse participates in that topic. He mentioned that only a subset of his efforts for integrating good Python compatibility in giac have reached the Prime's OS yet, and recommends testing giac's Python compat syntax code in xcas instead.

[critor]

Thanks to Jean Baptiste Boric, the NumWorks firmware is now available for your HP Prime calculator :
https://tiplanet.org/forum/gallery/image.php?album_id=137&image_id=9606

It works on hardware revision C.
The keypad doesn't work on my DVT prototype which is similar to hardware revision A. So it may not work on hardware revision A for now.

Alterations to the source code : https://github.com/numworks/epsilon/pull/534/commits

Binary download : https://tiplanet.org/forum/viewtopic.php...20#p232267

How to install the NumWorks firmware :
- keep a copy of your Documents/HP Connectivity Kit/Firmware/PRIME_OS.ROM file
- overwrite it with the downloaded file
- connect your calculator
- force a firmware update from the connectivity kit

How to reinstall your HP firmware :
- restore your original Documents/HP Connectivity Kit/Firmware/PRIME_OS.ROM file
- connect your calculator
- while holding the calculator [Symb] key, press and release the reset button on the back
- confirm the firmware installation in the connectivity kit

More information :
https://github.com/boricj/epsilon/issues/1
https://tiplanet.org/forum/viewtopic.php...86#p232286 (in french)
https://tiplanet.org/forum/viewtopic.php...20#p232062 (in french)

[JDW]

(06-05-2018 08:12 PM)critor Wrote:  Thanks to Jean Baptiste Boric, the NumWorks firmware is now available for your HP Prime calculator...

What's the point of running NumWorks firmware on a Prime unless that firmware makes the Prime a better calculator? Does it?

[Sadsilence]

[/quote]

What's the point of running NumWorks firmware on a Prime unless that firmware makes the Prime a better calculator? Does it?
[/quote]

What's the point of climbing up a 5000+, 6000+ or 7000+ metres high "hill", where you cannot live in a long term anyway? The hill is there, it is a challenge und it is fun (to some people). Any more reasons necessary? No!

[grsbanks]

(07-07-2018 10:30 PM)JDW Wrote:  What's the point of running NumWorks firmware on a Prime unless that firmware makes the Prime a better calculator? Does it?

Believe it or not, calculators are more than just tools to get a job done for the majority of people who populate this forum or at least who participate on it. This is no different to people feeling a totally irrational attachment to their car, motorcycle, Hi-Fi system or whatever.

Many of the guys here collect calculators, I'm one. This is no different to collecting, say, stamps. What is the point of collecting old stamps that can't even be used to send a letter? There is no practical reason to it. Do I need nearly 150 calculators to do my job? Certainly not. Is there a practical point to running NumWorks firmware on a Prime? Nope, but it's an interesting exercise.